release: v1.3.4

- refactor: 分页 KV cache（PagedCache+CacheView）替换固定 slot，删除 PrefixCache
- refactor: 推理引擎控制逻辑重写，修复连续批处理核心缺陷、线程安全问题
- refactor: KV 缓存槽位下沉到注意力层，移除 _remap_kv / _writeback_kv
- refactor: 统一采样路径为 SamplingPipeline batch tensor，删除 apply_sampling_strategies
- refactor: 设计模式优化 inference 模块导入结构（cache/sampling 独立）
- feat: 推理引擎前缀缓存（KV cache 复用）
- feat: OpenAI 兼容 chat completion API（流式+非流式+usage）
- feat: Anthropic 兼容 /v1/messages API，移除旧版 /generate 端点
- feat: GRPO CLI 接入 + on-policy，OpenAI API top_k 参数化
- feat: Checkpoint 支持 extra 通用扩展数据
- feat: Docker Compose 一键部署（GPU/CPU 双模式）
- feat: GRPO 训练参数补充，批处理训练参数表
- fix: 调度器延迟优化 — 移除 5ms 睡眠，修复 refill 任务丢失
- fix: CLI 参数缺失/重复、device_ids 越界、generate 参数名不一致
- fix: 长对话截断方向错误，保留最新 token 而非最早
- fix: remove_task 未释放 KV cache slot 导致第二轮对话死锁
- fix: KV cache 槽位索引错位、版本校验缺失、注意力掩码
- fix: scheduler 越界 bug，SchedulerCallback 回调阶段修正
- perf: _Result 改用 Condition.wait_for 消除非流式 CPU 空转
- perf: decode 每步张量预分配；input_ids 改用一次构建代替逐元素赋值
- refactor: 移除 device_ids 参数，统一 CUDA_VISIBLE_DEVICES
- docs: 更新文档以匹配分页 KV cache 等代码重构
- docs: 修正多处文档错误、补充训练参数说明

.gitignore

@@ -15,6 +15,7 @@
 !/.gitattributes
 !/.dockerignore
 !/Dockerfile
+!/docker-compose.yml
 !/assets/**
 !/CONTRIBUTING.md
 !/LICENSE

README.md

@@ -27,9 +27,6 @@
 
 ## 📖 Table of Contents
 
-<details open>
-<summary><b>English</b></summary>
-
 - [Features](#features)
 - [Quick Start](#quick-start)
 - [Documentation](#documentation)
@@ -37,8 +34,6 @@
 - [Community](#community)
 - [License](#license)
 
-</details>
-
 ---
 
 <a id="english"></a>
@@ -51,7 +46,8 @@
 - 💡 **Easy to Use**: Simple API with comprehensive examples and demos.
 - 📦 **Lightweight**: Minimal dependencies, easy to deploy.
 - 🔬 **Research‑Friendly**: Modular design, easy to experiment with new ideas.
-- 🤗 **HuggingFace Integration**: Compatible with HuggingFace models and datasets.
+- 🤗 **HuggingFace-Style API**: AutoModel/AutoTokenizer APIs inspired by HuggingFace for easy model and tokenizer loading.
+- 🔌 **Dual API Compatibility**: Supports both OpenAI and Anthropic chat completion APIs out of the box.
 
 ### Quick Start
 
@@ -72,16 +68,26 @@ pip install -e ".[dev]"
 #### Train a Model
 
 ```bash
-python scripts/tools/train.py \
-  --train_type=seq \
-  --data_root_path=/path/to/dataset \
-  --param_path=/path/to/param_path
+CUDA_VISIBLE_DEVICES=0,1,2,3 python scripts/tools/train.py \
+    --train_type seq \
+    --data_root_path /path/to/dataset \
+    --param_path /path/to/model \
+    --batch_size 4 \
+    --accumulation_steps 8 \
+    --max_lr 3e-4 \
+    --warmup_steps 1000 \
+    --n_epoch 1
 ```
 
+Full reference at [Parameter Guide](assets/docs/params.md).
+
 #### Generate Text
 
 ```bash
-python scripts/tools/generate.py --param_path=/path/to/param_path
+python scripts/tools/generate.py \
+    --param_path /path/to/model \
+    --input_json_file /path/to/input.json \
+    --output_json_file /path/to/output.json
 ```
 
 #### Docker
@@ -104,13 +110,19 @@ docker run --gpus all -p 8000:8000 astrai:latest \
 
 # Run with volume mount for data
 docker run --gpus all -v /path/to/data:/data -it astrai:latest
+
+# Docker Compose (GPU, default)
+docker compose up -d
+
+# Docker Compose (CPU only)
+docker compose --profile cpu up -d
 ```
 
 > **Note**: `--gpus all` is required for CUDA support. Without it, `torch.cuda.is_available()` will return `False`.
 
 #### Start HTTP Server
 
-Start the inference server with OpenAI-compatible HTTP API:
+Start the inference server with OpenAI and Anthropic-compatible HTTP API:
 
 ```bash
 python -m scripts.tools.server --port 8000 --device cuda
@@ -119,7 +131,7 @@ python -m scripts.tools.server --port 8000 --device cuda
 Make requests:
 
 ```bash
-# Chat API (OpenAI compatible)
+# OpenAI-compatible
 curl -X POST http://localhost:8000/v1/chat/completions \
   -H "Content-Type: application/json" \
   -d '{
@@ -127,7 +139,7 @@ curl -X POST http://localhost:8000/v1/chat/completions \
     "max_tokens": 512
   }'
 
-# Streaming response
+# OpenAI-compatible streaming
 curl -X POST http://localhost:8000/v1/chat/completions \
   -H "Content-Type: application/json" \
   -d '{
@@ -136,6 +148,27 @@ curl -X POST http://localhost:8000/v1/chat/completions \
     "max_tokens": 500
   }'
 
+# Anthropic-compatible
+curl -X POST http://localhost:8000/v1/messages \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "astrai",
+    "system": "You are a helpful assistant.",
+    "messages": [{"role": "user", "content": "Hello"}],
+    "max_tokens": 512
+  }'
+
+# Anthropic-compatible streaming with stop sequences
+curl -X POST http://localhost:8000/v1/messages \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "astrai",
+    "messages": [{"role": "user", "content": "Write a story"}],
+    "max_tokens": 500,
+    "stream": true,
+    "stop_sequences": ["The end"]
+  }'
+
 # Health check
 curl http://localhost:8000/health
 ```

assets/docs/README-zh-CN.md

@@ -52,7 +52,8 @@
 - 💡 **易用**: 简洁的 API 与丰富的示例、演示。
 - 📦 **轻量**: 依赖少，部署简单。
 - 🔬 **研究友好**: 模块化设计，便于实验新想法。
-- 🤗 **HuggingFace 集成**: 兼容 HuggingFace 模型与数据集。
+- 🤗 **HuggingFace 风格 API**: 类 HuggingFace 的 AutoModel/AutoTokenizer 接口，方便加载模型和分词器。
+- 🔌 **双 API 兼容**: 同时支持 OpenAI 和 Anthropic 聊天补全 API，开箱即用。
 
 ### 快速开始
 
@@ -73,16 +74,26 @@ pip install -e ".[dev]"
 #### 训练模型
 
 ```bash
-python scripts/tools/train.py \
-  --train_type=seq \
-  --data_root_path=/path/to/dataset \
-  --param_path=/path/to/param_path
+CUDA_VISIBLE_DEVICES=0,1,2,3 python scripts/tools/train.py \
+    --train_type seq \
+    --data_root_path /path/to/dataset \
+    --param_path /path/to/model \
+    --batch_size 4 \
+    --accumulation_steps 8 \
+    --max_lr 3e-4 \
+    --warmup_steps 1000 \
+    --n_epoch 1
 ```
 
+完整参数列表见[参数说明](./params.md)。
+
 #### 文本生成
 
 ```bash
-python scripts/tools/generate.py --param_path=/path/to/param_path
+python scripts/tools/generate.py \
+    --param_path /path/to/model \
+    --input_json_file /path/to/input.json \
+    --output_json_file /path/to/output.json
 ```
 
 #### Docker
@@ -105,13 +116,19 @@ docker run --gpus all -p 8000:8000 astrai:latest \
 
 # 挂载数据卷
 docker run --gpus all -v /path/to/data:/data -it astrai:latest
+
+# Docker Compose（GPU，默认）
+docker compose up -d
+
+# Docker Compose（仅 CPU）
+docker compose --profile cpu up -d
 ```
 
 > **注意**: 必须使用 `--gpus all` 才能启用 CUDA 支持，否则 `torch.cuda.is_available()` 将返回 `False`。
 
 #### 启动 HTTP 服务
 
-启动推理服务器，支持 OpenAI 兼容的 HTTP API：
+启动推理服务器，支持 OpenAI 和 Anthropic 兼容的 HTTP API：
 
 ```bash
 python -m scripts.tools.server --port 8000 --device cuda
@@ -120,7 +137,7 @@ python -m scripts.tools.server --port 8000 --device cuda
 发起请求：
 
 ```bash
-# Chat API（OpenAI 兼容）
+# OpenAI 兼容
 curl -X POST http://localhost:8000/v1/chat/completions \
   -H "Content-Type: application/json" \
   -d '{
@@ -128,7 +145,7 @@ curl -X POST http://localhost:8000/v1/chat/completions \
     "max_tokens": 512
   }'
 
-# 流式响应
+# OpenAI 兼容流式
 curl -X POST http://localhost:8000/v1/chat/completions \
   -H "Content-Type: application/json" \
   -d '{
@@ -137,6 +154,27 @@ curl -X POST http://localhost:8000/v1/chat/completions \
     "max_tokens": 500
   }'
 
+# Anthropic 兼容
+curl -X POST http://localhost:8000/v1/messages \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "astrai",
+    "system": "你是一个乐于助人的助手。",
+    "messages": [{"role": "user", "content": "你好"}],
+    "max_tokens": 512
+  }'
+
+# Anthropic 兼容流式并设置停止序列
+curl -X POST http://localhost:8000/v1/messages \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "astrai",
+    "messages": [{"role": "user", "content": "写个故事"}],
+    "max_tokens": 500,
+    "stream": true,
+    "stop_sequences": ["结束"]
+  }'
+
 # 健康检查
 curl http://localhost:8000/health
 ```

assets/docs/dataflow.md

@@ -7,14 +7,12 @@ This document describes the data flow of the AstrAI project (a training and infe
 AstrAI adopts a modular design with the following main components:
 - **Dataset Module** (`astrai/dataset/`): Dataset, sampler, serialization tools
 - **Model Module** (`astrai/model/`): AutoModel, Transformer model and its submodules
-- **Training Module** (`astrai/trainer/`): Trainer, training context, strategies, schedulers
+- **Training Module** (`astrai/trainer/`): Trainer, training context, strategies, schedulers, callbacks, metric utilities
 - **Inference Module** (`astrai/inference/`): Inference engine with continuous batching, streaming generation
-- **Config Module** (`astrai/config/`): Model, training, scheduler, and other configurations
+- **Config Module** (`astrai/config/`): ModelConfig, TrainConfig
 - **Factory Module** (`astrai/factory/`): Registry, BaseFactory for component registration
 - **Parallel Module** (`astrai/parallel/`): Distributed training support
-- **Serialization Module** (`astrai/serialization/`): HDF5 data loading, checkpoint management
-
-The data flow can generally be divided into two main lines: **Training Data Flow** and **Inference Data Flow**.
+- **Serialization** (`astrai/serialization.py`): HDF5 data loading, checkpoint management
 
 ## Data Flow Diagram
 
@@ -23,38 +21,36 @@ flowchart LR
     subgraph A[Data Preparation]
         direction TB
         A1[Raw Text] --> A2[AutoTokenizer]
-        A2 --> A3[Serialize to .h5 files]
+        A2 --> A3[Tokenized .h5 files]
         A3 --> A4[BaseDataset]
         A4 --> A5[ResumableDistributedSampler]
-        A5 --> A6[PyTorch DataLoader]
+        A5 --> A6[DataLoader]
     end
 
     subgraph B[Training]
         direction TB
-        B1[Batch Data] --> B2[TrainContextBuilder]
-        B2 --> B3[TrainContext]
-        B3 --> B4[BaseStrategy]
-        B4 --> B5[Transformer]
-        B5 --> B6[Compute Loss]
-        B6 --> B7[Backward]
-        B7 --> B8[Optimizer]
-        B8 --> B9[LRScheduler]
-        B9 --> B10[CheckpointCallback]
+        B1[DataLoader] --> B2[BaseStrategy]
+        B2 --> B3[Transformer Forward]
+        B3 --> B4[Loss + Backward]
+        B4 --> B5[Gradient Accumulation]
+        B5 -->|every accum_steps| B6[Optimizer Step]
+        B6 --> B7[LR Scheduler]
+        B7 -->|next batch| B2
+        B6 --> B8[CheckpointCallback]
     end
 
     subgraph C[Inference]
         direction TB
         C1[Checkpoint] --> C2[AutoModel]
-        C2 --> C3[Transformer + Tokenizer]
-        C3 --> C4[GenerationRequest + apply_chat_template]
-        C4 --> C5[InferenceEngine]
-        C5 --> C6[InferenceScheduler]
-        C6 --> C7[apply_sampling_strategies]
-        C7 --> C8[Transformer Forward]
-        C8 --> C9[KV Cache + Prefix Cache]
-        C9 --> C10{End Condition?}
-        C10 -->|No| C8
-        C10 -->|Yes| C11[Output Text]
+        C1 --> C3[AutoTokenizer]
+        C2 --> C4[InferenceEngine]
+        C3 --> C4
+        C4 --> C5[InferenceScheduler]
+        C5 --> C6[Transformer Forward]
+        C6 --> C7[sample]
+        C7 --> C8{End?}
+        C8 -->|No| C6
+        C8 -->|Yes| C9[Generated Text]
     end
 
     A --> B
@@ -63,207 +59,179 @@ flowchart LR
 
 ## Detailed Module Descriptions
 
-### 1. Dataset Module
+### 1. Serialization (`astrai/serialization.py`)
 
-#### 1.1 Serialization (`serialization.py`)
-- **`save_h5`**: Saves multiple tensors by groups as HDF5 files (`.h5`), each key corresponds to a list of tensors
-- **`load_h5`**: Loads `.h5` files, returns `Dict[str, List[Tensor]]`, supports shared memory (`share_memory=True`)
-- **`Checkpoint` class**: Encapsulates model state dict, training epoch, iteration count; supports safetensors format for saving and loading
+- **`save_h5`**: Saves tensors by groups as HDF5 files (`.h5`), each key maps to a list of tensors
+- **`load_h5`**: Loads `.h5` files, returns `Dict[str, List[Tensor]]`, supports shared memory
+- **`Checkpoint`**: Encapsulates model state dict + epoch + iteration; uses safetensors
 
-#### 1.2 Dataset (`dataset.py`)
-- **`BaseDataset`**: Abstract base class, defines common logic for window sampling, stride, etc.
-- **`BaseSegmentFetcher`** and **`MultiSegmentFetcher`**: Efficiently fetch data from specified index ranges in multiple segments
-- **`DatasetFactory`**: Factory pattern, supports dynamic registration of dataset types (`seq`, `sft`, `dpo`, `grpo`)
-- After dataset loading, multiple data keys (such as `"sequence"`, `"mask"`) are managed through `MultiSegmentFetcher`
+### 2. Dataset Module
 
-#### 1.3 Sampler (`sampler.py`)
-- **`ResumableDistributedSampler`**: Resumable sampler supporting distributed training
-- Records current epoch and iteration position, enabling training resume from breakpoints
-- Supports shuffle and drop_last options
+#### 2.1 Dataset (`dataset.py`)
+- **`BaseDataset`**: Abstract base class for windowed sequence sampling
+- **`BaseSegmentFetcher` / `MultiSegmentFetcher`**: Fetch tensor segments by index range
+- **`DatasetFactory`**: Creates dataset instances by `train_type` (`seq`, `sft`, `dpo`, `grpo`)
+- Data keys: `"sequence"` (SEQ), `"loss_mask"` (SFT), `"chosen_mask"/"rejected_mask"` (DPO), `"masks"` (GRPO)
 
-### 2. Model Module
+#### 2.2 Sampler (`sampler.py`)
+- **`ResumableDistributedSampler`**: Tracks `epoch` and `iter` for breakpoint resume; supports shuffle and drop_last
 
-#### 2.1 Transformer / AutoModel (`transformer.py`, `automodel.py`)
-- **`AutoModel`**: Base class for autoregressive language models with `from_pretrained()` and `save_pretrained()` methods
-- **`Transformer`**: Core autoregressive decoder architecture (registered via `@AutoModel.register('transformer')`)
-- Contains embedding layer, multi-layer `DecoderBlock`, RMSNorm, and linear output head
-- Supports weight tying (`tie_weight=True`) to reduce parameter count
-- Uses Rotary Position Embedding (RoPE) to inject position information
-- Supports loading from safetensors format with automatic model type detection from `config.json`
+### 3. Model Module
 
-#### 2.2 Submodules (`module.py`)
-- **`RotaryEmbedding`**: Generates RoPE cos/sin cache
-- **`DecoderBlock`**: Contains multi-head attention (supports GQA and MLA), feedforward network (FFN), residual connections
-- **`GQA`**: Grouped Query Attention implementation
-- **`MLA`**: Multi-Latent Attention implementation (like Qwen2-VL)
-- **`MLP`**: Feed-forward network with SiLU activation and gated mechanism
-- **`RMSNorm`**: Layer normalization variant
-- **`Linear`**, **`Embedding`**: Custom linear layer and embedding layer, supporting parallelism wrappers
+#### 3.1 Transformer / AutoModel
+- **`AutoModel`**: Base class with `from_pretrained()` / `save_pretrained()`
+- **`Transformer`**: Decoder-only architecture, registered via `@AutoModel.register('transformer')`
+- Embedding → N×DecoderBlock → RMSNorm → Linear lm_head
+- RoPE position encoding, optional weight tying
 
-### 3. Training Module
+#### 3.2 Submodules (`module.py`)
+- **`DecoderBlock`**: GQA attention + residual + MLP + RMSNorm
+- **`GQA`**: Grouped Query Attention (also `MLA` for multi-latent attention)
+- **`MLP`**: `SiLU(gate(x)) * up(x)` → down projection
+- **`RotaryEmbedding`**: RoPE cos/sin cache
+- **`RMSNorm`**: Layer normalization
 
-#### 3.1 Training Context (`train_context.py`)
-- **`TrainContext`**: Data class encapsulating all components needed for training (model, optimizer, data loader, strategy, etc.)
-- **`TrainContextBuilder`**: Builder pattern, progressively assembles training context, supports resume from checkpoint
+### 4. Training Module
 
-#### 3.2 Trainer (`trainer.py`)
-- **`Trainer`**: Main training loop, manages callbacks (progress bar, checkpoint, metric logging, gradient clipping, scheduler)
-- Supports distributed training (launches multi-process via `spawn_parallel_fn`)
-- Training steps include:
-  1. `on_train_begin` → 2. `on_epoch_begin` → 3. `on_batch_begin` → 4. Forward/loss calculation → 5. `on_batch_end` → 6. Gradient accumulation → 7. `on_step_begin` → 8. Optimizer update → 9. `on_step_end` → 10. `on_epoch_end`
+#### 4.1 Training Context (`train_context.py`)
+- **`TrainContext`**: Dataclass holding model, optimizer, dataloader, strategy, scheduler, checkpoint state
+- **`TrainContextBuilder`**: Builder pattern — takes checkpoint for resume, builds all components
 
-#### 3.3 Strategy (`strategy.py`)
-- **`BaseStrategy`**: Defines training strategy interface
-- **`SEQStrategy`**: Standard next-token prediction training
-- **`SFTStrategy`**: Supervised Fine-tuning with loss masking
-- **`DPOStrategy`**: Direct Preference Optimization
-- **`GRPOStrategy`**: Group Relative Policy Optimization
-- Strategy receives batch data, executes model forward pass, loss calculation, returns loss tensor
-- Created dynamically by `StrategyFactory` according to configuration
+#### 4.2 Trainer (`trainer.py`)
 
-#### 3.4 Scheduler (`schedule.py`)
-- **`BaseScheduler`**: Abstract base class defining learning rate scheduling interface
-- **`CosineScheduler`**: Cosine decay scheduler with warmup
-- **`SGDRScheduler`**: Stochastic Gradient Descent with Warm Restarts
-- **`SchedulerFactory`**: Factory pattern, supports registration of various schedulers
-- Scheduler is automatically created according to configuration and bound to optimizer
+The training loop is nested: **epoch** → **batch** (with step phase interspersed):
 
-#### 3.5 Callbacks (`train_callback.py`)
-- **`TrainCallback`**: Protocol interface for trainer callbacks
-- **`CheckpointCallback`**: Saves model checkpoints at configurable intervals
-- **`ProgressBarCallback`**: Displays training progress
-- **`MetricLoggerCallback`**: Logs training metrics to JSON files
-- **`GradientClippingCallback`**: Clips gradient norms
-- **`SchedulerCallback`**: Steps learning rate scheduler
+```
+on_train_begin
+  on_epoch_begin
+    for each batch:
+      if iteration % accumulation_steps == 0:        ← step phase
+        on_step_begin → optimizer.step() → zero_grad → on_step_end
+                                                      ← batch phase
+      on_batch_begin → strategy(batch) → loss → backward → on_batch_end
+      iteration += 1
 
-### 4. Factory Module
+    on_epoch_end
+on_train_end
+```
 
-#### 4.1 Registry and BaseFactory (`factory.py`)
-- **`Registry`**: Flexible registry for component classes with category and priority support
-- **`BaseFactory`**: Generic factory class for component registration and creation
-- Supports decorator-based registration pattern for extensible components
-- Provides methods for registration, retrieval, and listing with filtering
+Key points:
+- `on_step_*` wraps optimizer step (fires every `accumulation_steps` batches)
+- `on_batch_*` wraps loss computation (fires every batch)
+- `SchedulerCallback` fires on `on_batch_end` — LR scheduler steps every batch
+- `GradientClippingCallback` fires on `on_step_begin`
 
-### 5. Parallel Module
+#### 4.3 Strategy (`strategy.py`)
+- **`SEQStrategy`**: Next-token prediction, cross-entropy with label smoothing
+- **`SFTStrategy`**: Supervised fine-tuning with loss masking
+- **`DPOStrategy`**: Direct Preference Optimization with reference model
+- **`GRPOStrategy`**: Group Relative Policy Optimization with clipped ratio
 
-#### 5.1 Setup (`setup.py`)
-- **`spawn_parallel_fn`**: Spawns multiple processes for distributed training using PyTorch multiprocessing
-- **`setup_parallel`**: Context manager for initializing distributed process group (NCCL/CCL backend)
-- **`only_on_rank`**: Decorator to execute functions only on specific ranks
-- **`get_rank`**: Returns current process rank in distributed group
-- **`get_world_size`**: Returns total number of processes in distributed group
-- **`get_current_device`**: Returns current device from environment
+#### 4.4 Scheduler (`schedule.py`)
+- **`CosineScheduler`**: Cosine decay + linear warmup
+- **`SGDRScheduler`**: Cosine annealing with warm restarts
+- Created by `SchedulerFactory` and bound to optimizer
 
-#### 5.2 Parallel Layers (`module.py`)
-- **`ParallelModel`**: Base class for parallel models with process group
-- **`ColumnParallelLinear`**: Column-parallel linear layer with input splitting and output gathering
-- **`RowParallelLinear`**: Row-parallel linear layer with output reduction
+#### 4.5 Callbacks
+- **`CheckpointCallback`**: Saves safetensors at `ckpt_interval` iterations
+- **`ProgressBarCallback`**: tqdm progress display
+- **`MetricLoggerCallback`**: Writes JSONL metrics to `{ckpt_dir}/logs/`
+- **`GradientClippingCallback`**: `clip_grad_norm_` on `on_step_begin`
+- **`SchedulerCallback`**: `scheduler.step()` on `on_batch_end`
 
-### 6. Inference Module
+### 5. Inference Module
 
-#### 6.1 Inference Engine (`engine.py`)
-- **`InferenceEngine`**: Unified inference interface, supports streaming and non-streaming generation
-- **`InferenceScheduler`**: Continuous batching scheduler with dynamic batch composition
-- **`GenerationRequest`**: Encapsulates generation parameters (top_k, top_p, temperature, max_len, messages, etc.)
-- **`messages` format**: List of message dictionaries with `role` (system/user/assistant) and `content`
-- **`apply_chat_template`** (from `tokenizer.py`): Converts messages into prompt string using ChatML format
-- Provides streaming (`stream=True`) and non-streaming (`stream=False`) generation interfaces
-- Supports continuous batching with `max_batch_size` and `max_seq_len` parameters
-- Uses separate model and tokenizer initialization for flexibility
+#### 5.1 Inference Engine (`engine.py`)
+- **`InferenceEngine`**: Facade over scheduler; provides `generate()`, `generate_with_request()`, `generate_async()`
+- Accepts `prompt: str | List[str]`, returns generator (stream) or string (non-stream)
 
-#### 6.2 Scheduler (`scheduler.py`)
-- **`Task`**: Individual generation task with state management (PENDING, RUNNING, FINISHED, ABORTED)
-- **`TaskStatus`**: Task state enumeration
-- **`apply_sampling_strategies`**: Applies temperature, top-k, top-p sampling to logits
-- **`PrefixCacheManager`**: Radix tree-based prefix cache with LRU eviction for efficient KV cache reuse
-- **`RadixNode`**: Tree node structure for prefix caching
-- Continuous batching: new requests can join at any time, completed requests are released immediately
+#### 5.2 Scheduler 4-Phase Loop (`scheduler.py`)
 
-#### 6.3 Server (`server.py`)
-- FastAPI-based HTTP inference server
-- OpenAI-compatible `/v1/chat/completions` endpoint
-- Health check and statistics endpoints
-- Supports both streaming and non-streaming responses
+Background thread runs continuously:
 
-### 7. Tokenizer Module
+```
+1. Cleanup → Remove finished tasks, free KV cache pages
+2. Refill  → Pop from waiting_queue, alloc pages, add to active
+3. Prefill → Group active tasks by prompt_len, run full forward pass
+4. Decode  → Pick largest same-position group, run single-token forward
+```
 
-#### 7.1 Tokenizer (`tokenizer.py`)
-- Implemented based on HuggingFace tokenizers library (Byte-Level BPE)
-- **`AutoTokenizer`**: Auto-loading tokenizer class
-- Supports special tokens: `<｜begin▁of▁sentence｜>`, `<｜end▁of▁sentence｜>`, `<｜▁pad▁｜>`, `<｜im▁start｜>`, `<｜im▁end｜>`
-- Provides `encode`/`decode` methods for mutual conversion between text and token IDs
-- Uses `AutoTokenizer` for loading pre-trained tokenizers
+- **`Task`**: Tracks prompt_ids, output_ids, page_table, status (PENDING/RUNNING/FINISHED/ABORTED)
+- **`PagedCache`**: Bitmask-based page allocator with page-table-indirected read/write
+- **`CacheView`**: Batch view bundling cache + page table for attention layers
+- **`sample()`**: Temperature → top-k → top-p → multinomial
 
-#### 7.2 Chat Template (`chat_template.py`)
-- **`ChatTemplate`**: Jinja2-based chat template with rendering support
-- Handles multi-role message formatting (system, user, assistant)
-- Supports dynamic prompts and generation prompts
+#### 5.3 Server (`server.py`)
+- FastAPI with OpenAI `/v1/chat/completions` and Anthropic `/v1/messages` endpoints
+- Streaming via SSE, health check at `/health`, stats at `/stats`
 
-## Training Data Flow - Detailed Steps
+### 6. Tokenizer Module
+
+- **`AutoTokenizer`**: Wraps HuggingFace tokenizers (BBPE); `encode`/`decode`/`apply_chat_template`
+- **`ChatTemplate`**: Jinja2-based template rendering for multi-turn chat
+
+### 7. Factory & Parallel
+
+- **`Registry` / `BaseFactory`**: Decorator-based component registration
+- **`spawn_parallel_fn`**: Multi-process DDP launcher with NCCL backend
+- **`ParallelModel` / `ColumnParallelLinear` / `RowParallelLinear`**: Tensor model parallelism
+
+## Training Data Flow — Detailed Steps
 
 1. **Data Preparation**
-   - Raw text is converted to token ID sequences through AutoTokenizer
-   - Token ID sequences (possibly with masks, labels, etc.) are saved by groups as `.h5` files
-   - Files can contain multiple segments, each segment corresponds to a tensor
+   - Raw text → token IDs via `AutoTokenizer.encode()`
+   - Save as `.h5` files (groups of tensor lists per data key)
 
 2. **Dataset Loading**
-   - `BaseDataset`'s `load` method calls `load_h5`, obtaining `segments` dictionary
-   - Create `MultiSegmentFetcher` to manage data for multiple keys
-   - Calculate total sample count, and determine start/end indices for each sample based on window size and stride
+   - `BaseDataset.load()` calls `load_h5()`, builds `MultiSegmentFetcher`
+   - Sliding window of `window_size` with `stride` determines sample boundaries
 
-3. **Sampling and Batch Loading**
-   - `ResumableDistributedSampler` generates index sequence based on current epoch and iteration position
-   - PyTorch `DataLoader` uses sampler to get indices, calls dataset's `__getitem__` to get actual data
-   - Batch data shape is `[batch_size, window_size]` (or varies according to specific dataset type)
+3. **Sampling & Batching**
+   - `ResumableDistributedSampler` produces shuffled index sequences
+   - `DataLoader` fetches `[batch_size, window_size]` tensors via `__getitem__`
 
-4. **Strategy Forward and Loss Calculation**
-   - Batch data is passed to strategy (such as `SEQStrategy`)
-   - Strategy internally calls `Transformer` model, obtaining logits
-   - Calculate cross-entropy loss (or DPO loss, etc.) according to task type
-   - Return loss tensor
+4. **Strategy Forward**
+   - Strategy receives batch, calls `Transformer.forward()` for logits
+   - Computes task-specific loss (cross-entropy, DPO, GRPO)
 
-5. **Backpropagation and Optimization**
-   - Loss is normalized by dividing by accumulation steps, then `loss.backward()` is executed
-   - After accumulating `accumulation_steps` batches, optimizer `step()` and `zero_grad()` are executed
-   - Learning rate scheduler updates learning rate after each step
+5. **Backward & Accumulation**
+   - `loss = raw_loss / accumulation_steps`
+   - `loss.backward()` accumulates gradients
+   - Every `accumulation_steps` batches: `optimizer.step()` → `zero_grad()`
+   - Every batch: `scheduler.step()` updates learning rate
 
-6. **Checkpoint Saving**
-   - `CheckpointCallback` saves checkpoints at set intervals
-   - Checkpoints contain model state dict, current epoch, iteration, and other metadata
-   - Saved in safetensors format, ensuring safety and efficiency
+6. **Checkpoint**
+   - `CheckpointCallback` saves `model.state_dict()` + metadata to safetensors at `ckpt_interval` iterations
+   - Does NOT save optimizer/scheduler state (resume resets those)
 
-## Inference Data Flow - Detailed Steps
+## Inference Data Flow — Detailed Steps
 
 1. **Model Loading**
-   - Load `Transformer` model from checkpoint via `AutoModel.from_pretrained()`
-   - Set model to evaluation mode (`model.eval()`), enable inference mode (`torch.inference_mode`)
+   - `AutoModel.from_pretrained(path)` loads weights from safetensors
+   - `torch.inference_mode()` wraps generation
 
-2. **Prompt Construction and Encoding**
-   - User messages (list of dict with role and content) are converted to ChatML format string through `apply_chat_template` method in tokenizer
-   - Tokenizer encodes prompt string to token ID sequence `input_ids`
-   - For batch generation, use `pad_sequence` for padding
+2. **Prompt Construction**
+   - Messages → `apply_chat_template(messages, tokenize=False)` → prompt string
+   - `tokenizer.encode(prompt)` → token IDs (truncated to `max_prompt_len`)
 
-3. **Autoregressive Generation Loop**
-   - Initialize KV cache (optional) and prefix cache
-   - Loop until generating `max_len` tokens or encountering stop token:
-     - Input current `input_ids` (or cached new token) to model, obtain `logits`
-     - Apply `apply_sampling_strategies` (temperature, top-k, top-p) to `logits`
-     - Sample next token ID from the processed distribution
-     - Append new token to `input_ids`, while updating KV cache
-     - For streaming generation, yield each token to caller immediately
+3. **Continuous Batching Loop**
+   - **Cleanup**: Finished tasks → `stream_callback(STOP)`, free KV pages
+   - **Refill**: Pop from waiting queue, `PagedCache.alloc_n()` for prompt pages
+   - **Prefill**: Group by prompt length, run full forward with `start_pos=0`
+   - **Decode**: Pick position group with most tasks, single-token forward:
+     - Model forward → `logits` → `sample()` → next token ID
+     - Append to `output_ids`, update `output_tokens`
+     - `_maybe_alloc_page()` grows page table as needed
+     - `stream_callback(token)` for streaming clients
 
-4. **Decoding and Output**
-   - Decode generated token ID sequence to text through tokenizer
-   - Remove special tokens, return plain text response
+4. **Output**
+   - `tokenizer.decode(output_ids)` → text
+   - Return to caller (streaming: token-by-token; non-streaming: complete string)
 
-## Checkpoint and Serialization
+## Checkpoint & Serialization
 
-- **Training Checkpoint**: Saves model parameters, optimizer state, scheduler state, current epoch and iteration
-- **Model Parameters**: Supports safetensors format, automatically handles special logic like weight tying during loading
-- **Dataset Serialization**: HDF5 format supports efficient random access and shared memory, suitable for large-scale pre-training data
+- **Training Checkpoint**: safetensors weights + epoch/iteration metadata. Optimizer/scheduler state is NOT persisted.
+- **Inference Loading**: `AutoModel.from_pretrained()` loads from the same safetensors format.
+- **Dataset Serialization**: HDF5 with shared memory support for large-scale pre-training data.
 
-## Summary
-
-The data flow design of AstrAI reflects the characteristics of modularity, extensibility, and resumability. The training data flow supports large-scale distributed training through chunk loading, resumable sampling, gradient accumulation, and other mechanisms; the inference data flow achieves efficient text generation using KV cache, prefix caching, and sampling strategies. Clear interfaces between modules facilitate customization and extension.
-
-> Document Update Time: 2026-04-09
+> Document Update Time: 2026-05-09

assets/docs/design.md

@@ -50,7 +50,6 @@ classDiagram
             +str master_port
             +Callable parallel_wrapper
             +Callable state_dict_fn
-            +List[int] device_ids
             +str device_type
             +dict extra_kwargs
             +validate()
@@ -85,8 +84,8 @@ classDiagram
         }
 
         class BaseSegmentFetcher {
-            +List~Tensor~ segments
-            +List~int~ cum_lengths
+            +List[Tensor] segments
+            +List[int] cum_lengths
             +int total_length
             +fetch_data(begin_idx, end_idx) Tensor
         }
@@ -99,8 +98,8 @@ classDiagram
         }
 
         class ResumableDistributedSampler {
-            +int start_epoch
-            +int start_iter
+            +int epoch
+            +int iter
         }
 
         class DatasetFactory {
@@ -109,7 +108,9 @@ classDiagram
             +create(train_type, window_size, stride) BaseDataset
             +load(train_type, load_path, window_size, stride) BaseDataset
         }
+    }
 
+    namespace serialization {
         class Checkpoint {
             +dict state_dict
             +int epoch
@@ -122,7 +123,7 @@ classDiagram
     namespace model {
         class AutoModel {
             +ModelConfig config
-            +Dict _registry
+            +Registry _registry
             +register(model_type) decorator
             +get_model_class(model_type) Type
             +from_pretrained(path, disable_random_init) nn.Module
@@ -137,7 +138,7 @@ classDiagram
             +ModuleList layers
             +RMSNorm norm
             +Linear lm_head
-            +forward(input_ids, input_mask, persistent_key_values, start_pos) Dict
+            +forward(input_ids, input_mask, paged_cache, start_pos) Dict
             +load_state_dict(state_dict)
             +state_dict()
         }
@@ -147,7 +148,7 @@ classDiagram
             +RMSNorm input_norm
             +MLP mlp
             +RMSNorm post_attention_norm
-            +forward(x, rotary_emb, attention_mask, kv_cache, start_pos) Tensor
+            +forward(x, rotary_emb, attention_mask, paged_cache, start_pos) Tensor
         }
 
         class GQA {
@@ -156,18 +157,20 @@ classDiagram
             +int head_dim
             +Linear q_proj, k_proj, v_proj, o_proj
             +RMSNorm q_norm, k_norm
-            +forward(x, rotary_emb, mask, kv_cache, start_pos) Tensor
+            +forward(x, rotary_emb, mask, paged_cache, start_pos) Tensor
         }
 
         class MLA {
             +int n_heads
             +int n_kv_heads
             +int head_dim
-            +Linear q_a_proj, q_b_proj, q_c_proj
-            +Linear kv_a_proj, kv_b_proj, kv_c_proj
+            +int kv_lora_rank
+            +int qk_nope_head_dim
+            +int qk_rope_head_dim
+            +Linear q_proj, kv_a_proj, kv_b_proj
             +Linear o_proj
-            +RMSNorm q_norm, k_norm
-            +forward(x, rotary_emb, mask, kv_cache, start_pos) Tensor
+            +RMSNorm kv_norm
+            +forward(x, rotary_emb, mask, paged_cache, start_pos) Tensor
         }
 
         class MLP {
@@ -191,7 +194,7 @@ classDiagram
             +int dim
             +int max_len
             +float base
-            +forward(x, start_pos) Tuple~Tensor, Tensor~
+            +forward(x, start_pos) Tuple[Tensor, Tensor]
         }
 
         class Embedding {
@@ -202,14 +205,14 @@ classDiagram
 
     namespace tokenize {
         class AutoTokenizer {
-            +List~str~ stop_ids
+            +List[int] stop_ids
             +int bos_id
             +int eos_id
             +int pad_id
             +vocab_size int
-            +encode(tokens, out_ids, add_special_tokens) List~int~
+            +encode(tokens, out_ids, add_special_tokens) List[int]
             +decode(tokens, skip_special_tokens) str
-            +apply_chat_template(messages, tokenize) Union~str, List[int]~
+            +apply_chat_template(messages, tokenize) Union[str, List[int]]
             +set_chat_template(template)
             +load(path)
             +from_pretrained(path) AutoTokenizer
@@ -218,7 +221,7 @@ classDiagram
 
         class ChatTemplate {
             +String template_str
-            +render(messages, add_generation_prompt) str
+            +render(messages, system_prompt, **extra_variables) str
             +from_string(template) ChatTemplate
         }
     }
@@ -228,7 +231,7 @@ classDiagram
             +Dict _entries
             +register(name, component_cls, category, priority)
             +get(name) Type
-            +list_names() List~str~
+            +list_names() List[str]
         }
 
         class BaseFactory {
@@ -242,10 +245,10 @@ classDiagram
     namespace trainer {
         class Trainer {
             +TrainConfig train_config
-            +List~TrainCallback~ callbacks
+            +List[TrainCallback] callbacks
             +train(checkpoint)
             +_build_context(checkpoint) TrainContext
-            +_get_default_callbacks() List~TrainCallback~
+            +_get_default_callbacks() List[TrainCallback]
         }
 
         class TrainContext {
@@ -265,8 +268,6 @@ classDiagram
         class TrainContextBuilder {
             +TrainConfig config
             +with_checkpoint(checkpoint) TrainContextBuilder
-            +with_dataloader() TrainContextBuilder
-            +with_strategy() TrainContextBuilder
             +build() TrainContext
         }
 
@@ -308,7 +309,7 @@ classDiagram
         }
 
         class BaseScheduler {
-            +get_lr() List~float~
+            +get_lr() List[float]
             +step()
         }
 
@@ -390,12 +391,9 @@ classDiagram
             +InferenceScheduler scheduler
             +int max_batch_size
             +Optional int max_seq_len
-            +int max_prefix_len
-            +int cache_capacity
-            +Tensor kv_cache
-            +Tensor seq_mask
             +generate(prompt, stream, max_tokens, temperature, top_p, top_k) Union[Generator, str, List[str]]
             +generate_with_request(request) Union[Generator, str, List[str]]
+            +generate_async(prompt, max_tokens, temperature, top_p, top_k) AsyncGenerator
             +get_stats() Dict
             +shutdown()
         }
@@ -403,10 +401,11 @@ classDiagram
         class InferenceScheduler {
             +nn.Module model
             +AutoTokenizer tokenizer
-            +ModelConfig config
-            +Tuple kv_cache
-            +Tensor seq_mask
-            +PrefixCacheManager prefix_cache
+            +PagedCache page_cache
+            +int max_batch_size
+            +int max_seq_len
+            +int max_prompt_len
+            +int page_size
             +List waiting_queue
             +List active_tasks
             +add_task(prompt, max_tokens, temperature, top_p, top_k, stream_callback) str
@@ -416,22 +415,26 @@ classDiagram
             +get_stats() Dict
         }
 
-        class PrefixCacheManager {
-            +RadixNode root
-            +int max_capacity
-            +List lru
-            +insert(token_ids, slot)
-            +find_longest_prefix(token_ids) Tuple[int, int]
-            +release(token_ids)
+        class PagedCache {
+            +int page_size
+            +int _free_mask
+            +List[int] _refs
+            +Tensor k_cache
+            +Tensor v_cache
+            +alloc() int
+            +alloc_n(n) List[int]
+            +free(idx)
+            +bind(page_table, total_len) CacheView
+            +write(layer_id, page_table, start_pos, k, v)
+            +gather(layer_id, page_table) Tuple[Tensor, Tensor]
         }
 
-        class RadixNode {
-            +Dict children
-            +int hash
-            +int slot
-            +int ref_count
-            +float last_access
-            +List token_sequence
+        class CacheView {
+            +PagedCache _cache
+            +Tensor _page_table
+            +int _total_len
+            +write(layer_id, start_pos, k, v)
+            +gather(layer_id) Tuple[Tensor, Tensor]
         }
 
         class Task {
@@ -445,38 +448,71 @@ classDiagram
             +List output_ids
             +int input_tokens
             +int output_tokens
-            +int slot
+            +List[int] page_table
+            +int n_pages
+            +float arrival_time
+            +float finish_time
             +Callable stream_callback
+            +int next_pos
             +is_finished(stop_ids) bool
         }
 
         class TaskStatus {
-            +str PENDING
-            +str RUNNING
-            +str FINISHED
-            +str ABORTED
-        }
-
-        class Server {
-            +start()
-            +predict(request)
+            <<enumeration>>
+            PENDING
+            RUNNING
+            FINISHED
+            ABORTED
         }
 
         class GenerationRequest {
+            +List[Dict] messages
+            +GenerationParams params
+            +bool stream
+        }
+
+        class GenerationParams {
+            <<value object>>
             +int top_k
             +float top_p
             +float temperature
-            +int max_len
-            +List~Dict~ messages
-            +stream bool
+            +int max_tokens
+        }
+
+        class BaseSamplingStrategy {
+            <<abstract>>
+            +apply(logits, filter_value) Tensor
+        }
+
+        class TemperatureStrategy {
+            +float temperature
+            +apply(logits, filter_value) Tensor
+        }
+
+        class TopKStrategy {
+            +int top_k
+            +apply(logits, filter_value) Tensor
+        }
+
+        class TopPStrategy {
+            +float top_p
+            +apply(logits, filter_value) Tensor
+        }
+
+        class SamplingPipeline {
+            +List strategies
+            +apply(logits, filter_value) Tensor
+            +sample(logits, filter_value) Tensor
         }
 
         class _Result {
-            +List~str~ tokens
-            +List~str~ results
-            +List~bool~ done_flags
+            +List[str] tokens
+            +List[str] results
+            +List[bool] _done
             +append(token, idx)
-            +get_results() List~str~
+            +get_results() List[str]
+            +pop_all() List[str]
+            +wait(timeout) bool
         }
 
         class ChatMessage {
@@ -485,28 +521,21 @@ classDiagram
         }
 
         class ChatCompletionRequest {
-            +List~ChatMessage~ messages
+            +List[ChatMessage] messages
             +float temperature
             +float top_p
             +int top_k
             +int max_tokens
             +bool stream
-            +Optional~str~ system_prompt
-        }
-
-        class CompletionResponse {
-            +str id
-            +str object
-            +int created
-            +str model
-            +List~Dict~ choices
+            +Optional[str] stop
+            +Optional[int] n
         }
     }
 
     namespace parallel {
-        class ParallelSetup {
+        class ParallelFunctions {
             +spawn_parallel_fn(fn, nprocs)
-            +setup_parallel(rank, world_size, backend, master_addr, master_port, device_type, device_ids)
+            +setup_parallel(rank, world_size, backend, master_addr, master_port, device_type)
         }
 
         class ParallelModel {
@@ -539,10 +568,10 @@ classDiagram
     Trainer --> TrainContextBuilder : builds
     Trainer --> TrainCallback : manages
     TrainContextBuilder --> TrainContext : creates
+    Checkpoint ..> Checkpoint : saves/loads
     TrainContext --> Checkpoint : manages
     TrainContext --> BaseStrategy : uses
     TrainContext --> BaseScheduler : uses
-    AutoModel --> ModelConfig : contains
     SchedulerFactory ..> BaseScheduler : creates
     BaseScheduler <|-- CosineScheduler
     BaseScheduler <|-- SGDRScheduler
@@ -553,30 +582,32 @@ classDiagram
     TrainCallback <|-- ProgressBarCallback
     TrainCallback <|-- MetricLoggerCallback
     InferenceEngine --> InferenceScheduler : uses
+    InferenceEngine --> GenerationRequest : uses
+    GenerationRequest --> GenerationParams : contains
     InferenceScheduler --> Task : manages
+    Task --> TaskStatus : uses
     InferenceScheduler --> TaskStatus : uses
+    InferenceScheduler --> PagedCache : uses
     InferenceScheduler --> Transformer : uses
     InferenceEngine --> Transformer : uses
-    InferenceEngine --> GenerationRequest : uses
-    Server --> InferenceEngine : uses
-    Server --> ChatMessage : uses
-    Server --> ChatCompletionRequest : uses
-    Server --> CompletionResponse : uses
-    ParallelSetup --> Trainer : enables
+    InferenceEngine --> _Result : uses
+    BaseSamplingStrategy <|-- TemperatureStrategy
+    BaseSamplingStrategy <|-- TopKStrategy
+    BaseSamplingStrategy <|-- TopPStrategy
+    SamplingPipeline --> BaseSamplingStrategy : composes
     BaseDataset <|-- SEQDataset
     BaseDataset <|-- SFTDataset
     BaseDataset <|-- DPODataset
     BaseDataset <|-- GRPODataset
     DatasetFactory ..> BaseDataset : creates
-    BaseSegmentFetcher --> MultiSegmentFetcher : used by
-    MultiSegmentFetcher --> BaseDataset : used by
+    MultiSegmentFetcher --> BaseSegmentFetcher : uses
+    BaseDataset --> MultiSegmentFetcher : uses
     AutoModel <|-- Transformer
     AutoModel --> ModelConfig : contains
     Transformer --> DecoderBlock : uses
     Transformer --> RotaryEmbedding : uses
     Transformer --> Embedding : uses
     DecoderBlock --> GQA : uses
-    DecoderBlock --> MLA : uses
     DecoderBlock --> MLP : uses
     DecoderBlock --> RMSNorm : uses
     TrainContextBuilder --> ResumableDistributedSampler : creates
@@ -584,9 +615,6 @@ classDiagram
     ParallelModel <|-- RowParallelLinear
     ParallelModel <|-- ColumnParallelLinear
     AutoTokenizer --> ChatTemplate : uses
-    InferenceScheduler --> PrefixCacheManager : uses
-    InferenceScheduler --> RadixNode : uses
-    Checkpoint ..> Checkpoint : saves/loads
     TrainConfig --> DatasetFactory : selects
     TrainConfig --> SchedulerFactory : selects
     TrainConfig --> CallbackFactory : selects
@@ -602,12 +630,13 @@ classDiagram
 | Module | Components | Description |
 |--------|------------|-------------|
 | **astrai.config** | ModelConfig, TrainConfig | Configuration management |
-| **astrai.dataset** | BaseDataset, SEQDataset, SFTDataset, DPODataset, GRPODataset, BaseSegmentFetcher, MultiSegmentFetcher, ResumableDistributedSampler, DatasetFactory, Checkpoint | Dataset loading and management |
+| **astrai.dataset** | BaseDataset, SEQDataset, SFTDataset, DPODataset, GRPODataset, BaseSegmentFetcher, MultiSegmentFetcher, ResumableDistributedSampler, DatasetFactory | Dataset loading and management |
+| **astrai.serialization** | Checkpoint, save_h5, load_h5 | Model serialization and checkpoint management |
 | **astrai.model** | AutoModel, Transformer, DecoderBlock, GQA, MLA, MLP, RMSNorm, Linear, RotaryEmbedding, Embedding | Neural network model |
 | **astrai.tokenize** | AutoTokenizer, ChatTemplate | Tokenizer and chat template |
 | **astrai.trainer** | Trainer, TrainContext, TrainContextBuilder, BaseStrategy, StrategyFactory, BaseScheduler, SchedulerFactory, TrainCallback, CallbackFactory | Training workflow management |
-| **astrai.inference** | InferenceEngine, InferenceScheduler, Task, TaskStatus, Server, GenerationRequest, PrefixCacheManager, ChatMessage, ChatCompletionRequest, CompletionResponse | Inference service with continuous batching |
-| **astrai.parallel** | ParallelSetup, ColumnParallelLinear, RowParallelLinear | Distributed parallel |
+| **astrai.inference** | InferenceEngine, InferenceScheduler, PagedCache, CacheView, Task, TaskStatus, GenerationParams, GenerationRequest, BaseSamplingStrategy, TemperatureStrategy, TopKStrategy, TopPStrategy, SamplingPipeline, ChatMessage, ChatCompletionRequest | Inference service with continuous batching and paged KV cache |
+| **astrai.parallel** | ParallelFunctions, ParallelModel, ColumnParallelLinear, RowParallelLinear | Distributed parallel |
 | **astrai.factory** | Registry, BaseFactory | Generic component registration |
 
 ### Design Patterns
@@ -618,8 +647,10 @@ classDiagram
 | **Builder** | `TrainContextBuilder` | Chain-building training context, step-by-step initialization of components |
 | **Factory** | `StrategyFactory`, `SchedulerFactory`, `DatasetFactory`, `CallbackFactory`, `BaseFactory` | Decorator registration mechanism, dynamically create training strategies, schedulers, datasets, and callbacks |
 | **Observer** | `TrainCallback`, `CallbackFactory` | Callback mechanism for training process monitoring (checkpoint, early stopping, metrics) |
-| **Singleton** | `TrainContext` | Training process global state management |
+| **Context** | `TrainContext` | Training process state container with model, optimizer, scheduler and checkpoint |
 | **Registry** | `BaseFactory`, `Registry` | Generic component registration with category and priority support |
+| **Object Pool** | `PagedCache` | Page-based KV cache with O(1) alloc/free via bitmask |
+| **Strategy (Sampling)** | `BaseSamplingStrategy`, `TemperatureStrategy`, `TopKStrategy`, `TopPStrategy`, `SamplingPipeline` | Composable logit transformations with temperature, top-k, top-p |
 | **Producer-Consumer** | `InferenceScheduler`, `Task`, `waiting_queue`, `active_tasks` | Continuous batching with dynamic task queue management |
 | **Event-Driven** | `threading.Event`, `_task_event` | Non-blocking wait mechanism for task scheduling using Python's `threading` module |
 | **AutoModel Registry** | `AutoModel`, `Transformer` | Model type registration and dynamic loading via decorator pattern |
@@ -630,8 +661,8 @@ classDiagram
 1. **Configuration → Training**: `TrainConfig` contains `ModelConfig`, holds model, dataset, optimizer and other references
 2. **Training Flow**: `Trainer` → `TrainContextBuilder` → `TrainContext`, uses `BaseStrategy` to compute loss
 3. **Strategy Selection**: `StrategyFactory` creates corresponding strategy instance based on `train_type`
-4. **Inference Flow**: `Server` → `InferenceEngine` → `InferenceScheduler` → `Transformer`, supports continuous batching with streaming/non-streaming
-5. **Distributed Support**: `ParallelSetup` provides multi-process training capability for `Trainer`
+4. **Inference Flow**: `InferenceEngine` → `InferenceScheduler` → `Transformer`, uses `PagedCache` for paged KV cache management and `SamplingPipeline` for efficient continuous batching with streaming/non-streaming
+5. **Distributed Support**: `spawn_parallel_fn` and `setup_parallel` provide multi-process training capability for `Trainer`
 6. **Dataset Loading**: `DatasetFactory` creates datasets (SEQDataset, SFTDataset, DPODataset, GRPODataset), supports HDF5 loading via `BaseSegmentFetcher` and `MultiSegmentFetcher`
 7. **Checkpoint Management**: `Checkpoint` handles model state serialization/deserialization with safetensors
 8. **Scheduler Support**: `SchedulerFactory` creates learning rate schedulers (CosineScheduler, SGDRScheduler)
@@ -675,12 +706,6 @@ $$
 L_{\text{GRPO}} = -\mathbb{E} \left[ \min\left( \frac{\pi_\theta(a|s)}{\pi_{\text{ref}}(a|s)} \cdot A, \text{clip}\left(\frac{\pi_\theta(a|s)}{\pi_{\text{ref}}(a|s)}, 1-\epsilon, 1+\epsilon\right) \cdot A \right) \right] + \lambda \cdot D_{KL}
 $$
 
-In this implementation, an off-policy approach is used ($\pi_\theta = \pi_{\text{ref}}$), and the policy loss simplifies to:
-
-$$
-L_{\text{policy}} = -\mathbb{E}[A]
-$$
-
 The KL divergence term uses mean squared error approximation:
 
 $$

assets/docs/introduction.md

@@ -2,7 +2,7 @@
 
 ### 1. Model Architecture
 
-This model uses the Transformer architecture with GQA mechanism (q_head=24, kv_head=4), which saves KV cache memory compared to traditional MHA. The model is built by stacking 32 layers of Transformer blocks, with 1.0 billion parameters. Transformer is an autoregressive model that calculates the relationship between all previous tokens to obtain the probability distribution of the next token.
+This model uses the Transformer architecture with GQA mechanism (q_head=24, kv_head=4), which saves KV cache memory compared to traditional MHA. The model is built by stacking 24 layers of Transformer blocks, with 1.0 billion parameters. Transformer is an autoregressive model that calculates the relationship between all previous tokens to obtain the probability distribution of the next token.
 
 The model now uses the **AutoModel** base class for flexible loading and saving:
 
@@ -48,14 +48,15 @@ flowchart TB
         S --> T[+]
         H --> T
         T --> U[RMSNorm]
-        U --> V[Linear]
-        V --> W[SiLU]
-        V --> X[×]
-        W --> X
-        X --> Y[Linear]
-        Y --> Z[+]
-        T --> Z
-        Z --> AA[x']
+        U --> V["Linear (gate)"]
+        U --> W["Linear (up)"]
+        V --> X[SiLU]
+        X --> Y[×]
+        W --> Y
+        Y --> Z["Linear (down)"]
+        Z --> AA[+]
+        T --> AA
+        AA --> BB[x']
     end
 
     classDef main fill:#e6f3ff,stroke:#0066cc;
@@ -168,8 +169,6 @@ from astrai.inference import InferenceEngine, GenerationRequest
 engine = InferenceEngine(
     model=model,
     tokenizer=tokenizer,
-    max_batch_size=8,
-    max_seq_len=4096,
 )
 
 # Use GenerationRequest with messages format
@@ -222,12 +221,11 @@ curl -X POST http://localhost:8000/v1/chat/completions \
 | Parameter | Type | Default | Description |
 |-----------|------|---------|-------------|
 | `messages` | List[dict] | Required | Chat messages with role and content |
-| `temperature` | float | 0.8 | Sampling temperature (0.0-2.0) |
-| `top_p` | float | 0.95 | Nucleus sampling threshold |
+| `temperature` | float | 1.0 | Sampling temperature (0.0-2.0) |
+| `top_p` | float | 1.0 | Nucleus sampling threshold |
 | `top_k` | int | 50 | Top-k sampling parameter |
-| `max_tokens` | int | 2048 | Maximum tokens to generate |
+| `max_tokens` | int | 1024 | Maximum tokens to generate |
 | `stream` | bool | false | Enable streaming response |
-| `system_prompt` | str | None | System prompt override |
 
 **Response (non-streaming):**
 ```json
@@ -242,7 +240,12 @@ curl -X POST http://localhost:8000/v1/chat/completions \
       "message": {"role": "assistant", "content": "Hello! I'm doing well..."},
       "finish_reason": "stop"
     }
-  ]
+  ],
+  "usage": {
+    "prompt_tokens": 20,
+    "completion_tokens": 15,
+    "total_tokens": 35
+  }
 }
 ```
 
@@ -262,25 +265,57 @@ curl -X POST http://localhost:8000/v1/chat/completions \
 
 The server uses Server-Sent Events (SSE) with content type `text/event-stream`.
 
-### Simple Generation Endpoint
+### Anthropic-Compatible Endpoint
 
-For basic text generation without chat format:
+The server also provides an Anthropic-compatible endpoint at `/v1/messages`:
 
 ```bash
-curl -X POST "http://localhost:8000/generate?query=Hello&max_len=1000" \
-  -H "Content-Type: application/json"
-```
-
-Or with conversation history:
-
-```bash
-curl -X POST "http://localhost:8000/generate" \
+curl -X POST http://localhost:8000/v1/messages \
   -H "Content-Type: application/json" \
   -d '{
-    "query": "What is AI?",
-    "history": [["Hello", "Hi there!"], ["How are you?", "I'm doing well"]],
-    "temperature": 0.8,
-    "max_len": 2048
+    "model": "astrai",
+    "system": "You are a helpful assistant.",
+    "messages": [{"role": "user", "content": "Hello, how are you?"}],
+    "max_tokens": 2048
+  }'
+```
+
+Response:
+```json
+{
+  "id": "msg_abc123...",
+  "type": "message",
+  "role": "assistant",
+  "model": "astrai",
+  "content": [{"type": "text", "text": "Hello! I am doing well..."}],
+  "stop_reason": "end_turn",
+  "stop_sequence": null,
+  "usage": {"input_tokens": 20, "output_tokens": 15}
+}
+```
+
+Streaming:
+```bash
+curl -X POST http://localhost:8000/v1/messages \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "astrai",
+    "system": "You are a helpful assistant.",
+    "messages": [{"role": "user", "content": "Write a short poem"}],
+    "max_tokens": 500,
+    "stream": true
+  }'
+```
+
+Supports `stop_sequences` for early termination:
+```bash
+curl -X POST http://localhost:8000/v1/messages \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "astrai",
+    "messages": [{"role": "user", "content": "Write a story"}],
+    "max_tokens": 500,
+    "stop_sequences": ["The end", "THE END"]
   }'
 ```
 
@@ -290,10 +325,10 @@ Monitor server and model status:
 
 ```bash
 curl http://localhost:8000/health
-# {"status": "ok", "model_loaded": true, "engine_ready": true}
+# {"status": "ok", "model_loaded": true}
 
 curl http://localhost:8000/stats
-# {"requests_total": 10, "tokens_generated": 5000, ...}
+# {"total_tasks": 10, "total_tokens": 5000, "active_tasks": 1, "waiting_queue": 0}
 ```
 
 > Document Update Time: 2026-04-09

assets/docs/params.md

@@ -4,70 +4,87 @@
 
 ### Basic Parameters
 
-| Parameter | Description | Default Value |
-|-----------|-------------|---------------|
-| `--train_type` | Training type (seq, sft, dpo, grpo) | required |
-| `--model_type` | Model type for AutoModel loading (e.g., transformer) | transformer |
+| Parameter | Description | Default |
+|-----------|-------------|---------|
+| `--train_type` | Training type (`seq`, `sft`, `dpo`, `grpo`) | required |
 | `--data_root_path` | Dataset root directory | required |
 | `--param_path` | Model parameters or checkpoint path | required |
 | `--n_epoch` | Total training epochs | 1 |
-| `--batch_size` | Batch size | 4 |
-| `--accumulation_steps` | Gradient accumulation steps | 1 |
+| `--batch_size` | Batch size | 1 |
+| `--accumulation_steps` | Gradient accumulation steps between optimizer steps | 1 |
 
 ### Learning Rate Scheduling
 
-| Parameter | Description | Default Value |
-|-----------|-------------|---------------|
+| Parameter | Description | Default |
+|-----------|-------------|---------|
 | `--warmup_steps` | Warmup steps | 1000 |
-| `--max_lr` | Maximum learning rate (warmup + cosine decay) | 3e-4 |
-| `--max_grad_norm` | Maximum gradient norm | 1.0 |
+| `--max_lr` | Maximum learning rate (cosine decay after warmup) | 3e-4 |
+| `--max_grad_norm` | Maximum gradient norm for clipping | 1.0 |
 
-### Checkpoint
+### Optimizer (AdamW)
 
-| Parameter | Description | Default Value |
-|-----------|-------------|---------------|
-| `--ckpt_interval` | Checkpoint save interval (iterations) | 5000 |
-| `--ckpt_dir` | Checkpoint save directory | checkpoint |
-| `--resume_dir` | Resume training from specified path | - |
-
-### Optimizer Parameters
-
-| Parameter | Description | Default Value |
-|-----------|-------------|---------------|
+| Parameter | Description | Default |
+|-----------|-------------|---------|
 | `--adamw_beta1` | AdamW beta1 | 0.9 |
 | `--adamw_beta2` | AdamW beta2 | 0.95 |
 | `--adamw_weight_decay` | AdamW weight decay | 0.01 |
 
 ### Data Loading
 
-| Parameter | Description | Default Value |
-|-----------|-------------|---------------|
-| `--random_seed` | Random seed | 3407 |
-| `--num_workers` | DataLoader workers | 0 |
-| `--prefetch_factor` | Prefetch factor for dataloader | None |
-| `--pin_memory` | Enable pin_memory | False |
-| `--no_pin_memory` | Disable pin_memory | - |
+| Parameter | Description | Default |
+|-----------|-------------|---------|
+| `--window_size` | Max input sequence length | model config `max_len` |
+| `--stride` | Stride for sliding window over sequences | None |
+| `--random_seed` | Random seed for reproducibility | 3407 |
+| `--num_workers` | DataLoader worker processes | 4 |
+| `--no_pin_memory` | Disable pin_memory (enabled by default) | (flag) |
+
+### Checkpoint & Resume
+
+| Parameter | Description | Default |
+|-----------|-------------|---------|
+| `--ckpt_interval` | Iterations between checkpoints | 5000 |
+| `--ckpt_dir` | Checkpoint save directory | checkpoint |
+| `--start_epoch` | Resume from epoch (0 = from scratch) | 0 |
+| `--start_batch` | Resume from batch iteration | 0 |
 
 ### Distributed Training
 
-| Parameter | Description | Default Value |
-|-----------|-------------|---------------|
-| `--nprocs` | Number of GPUs | 1 |
-| `--device_type` | Device type (cuda/cpu) | cuda |
+| Parameter | Description | Default |
+|-----------|-------------|---------|
+| `--nprocs` | Number of GPUs / processes | 1 |
+| `--device_type` | Device type | cuda |
 
-### Other Parameters
+### Strategy-specific
 
-| Parameter | Description | Default Value |
-|-----------|-------------|---------------|
-| `--window_size` | Maximum input sequence length | model config max_len |
-| `--stride` | Input sequence stride | - |
-| `--dpo_beta` | DPO beta value | 0.1 |
-| `--grpo_clip_eps` | GRPO clip epsilon | 0.2 |
-| `--grpo_kl_coef` | GRPO KL coefficient | 0.01 |
-| `--grpo_group_size` | GRPO group size | 4 |
-| `--label_smoothing` | Label smoothing parameter | 0.1 |
-| `--start_epoch` | Starting epoch | 0 |
-| `--start_batch` | Starting batch | 0 |
+| Parameter | Description | Default | Used by |
+|-----------|-------------|---------|---------|
+| `--dpo_beta` | DPO beta value | 0.1 | `dpo` |
+| `--label_smoothing` | Label smoothing for cross-entropy loss | 0.1 | `seq`, `sft` |
+| `--group_size` | GRPO group size | 4 | `grpo` |
+| `--grpo_clip_eps` | GRPO clipping epsilon | 0.2 | `grpo` |
+| `--grpo_kl_coef` | GRPO KL penalty coefficient | 0.01 | `grpo` |
+| `--grpo_sync_interval` | GRPO ref_model sync interval (steps) | 200 | `grpo` |
+
+### Usage Example
+
+```bash
+python scripts/tools/train.py \
+  --train_type seq \
+  --data_root_path /path/to/dataset \
+  --param_path /path/to/model \
+  --n_epoch 3 \
+  --batch_size 4 \
+  --accumulation_steps 8 \
+  --max_lr 3e-4 \
+  --warmup_steps 2000 \
+  --max_grad_norm 1.0 \
+  --ckpt_interval 5000 \
+  --ckpt_dir ./checkpoints \
+  --num_workers 4 \
+  --nprocs 1 \
+  --device_type cuda
+```
 
 ---
 
@@ -89,14 +106,14 @@
 ```python
 import torch
 from astrai.model import AutoModel
-from astrai.tokenize import Tokenizer
+from astrai.tokenize import AutoTokenizer
 from astrai.inference import InferenceEngine, GenerationRequest
 
 # Load model using AutoModel
 model = AutoModel.from_pretrained("your_model_dir")
 
 # Load tokenizer
-tokenizer = Tokenizer("your_model_dir")
+tokenizer = AutoTokenizer.from_pretrained("your_model_dir")
 
 # Create engine with separate model and tokenizer
 engine = InferenceEngine(

astrai/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "1.3.3"
+__version__ = "1.3.4"
 __author__ = "ViperEkura"
 
 from astrai.config import (

astrai/config/train_config.py

@@ -1,5 +1,5 @@
 from dataclasses import dataclass, field
-from typing import Callable, List, Optional
+from typing import Callable, Optional
 
 import torch.nn as nn
 from torch.optim import Optimizer
@@ -74,9 +74,6 @@ class TrainConfig:
     )
 
     # others
-    device_ids: Optional[List[int]] = field(
-        default=None, metadata={"help": "Device ids for distributed training."}
-    )
     device_type: str = field(
         default="cuda", metadata={"help": "Device type for distributed training."}
     )

astrai/inference/__init__.py

@@ -1,25 +1,46 @@
-"""Inference module for continuous batching."""
+"""Inference module for continuous batching.
+
+Layers:
+  - engine.py:    Facade (InferenceEngine), Value Object (GenerationParams, GenerationRequest)
+  - scheduler.py: Continuous-batching loop, Task state machine, TaskStatus enum
+  - cache.py:     PagedCache (page-table-indirected KV cache with alloc/free)
+  - sampling.py:  Strategy pattern (TemperatureStrategy, TopKStrategy, TopPStrategy)
+  - server.py:    FastAPI HTTP server (OpenAI-compatible endpoints)
+"""
 
 from astrai.inference.engine import (
+    GenerationParams,
     GenerationRequest,
     InferenceEngine,
 )
+from astrai.inference.sampling import (
+    BaseSamplingStrategy,
+    SamplingPipeline,
+    TemperatureStrategy,
+    TopKStrategy,
+    TopPStrategy,
+    sample,
+)
 from astrai.inference.scheduler import (
     InferenceScheduler,
     Task,
     TaskStatus,
-    apply_sampling_strategies,
 )
 
 __all__ = [
-    # Engine
+    # Engine / Requests
     "InferenceEngine",
+    "GenerationRequest",
+    "GenerationParams",
     # Scheduler
     "InferenceScheduler",
     "Task",
     "TaskStatus",
-    # Request
-    "GenerationRequest",
-    # Sampling
-    "apply_sampling_strategies",
+    # Sampling (Strategy pattern)
+    "sample",
+    "BaseSamplingStrategy",
+    "TemperatureStrategy",
+    "TopKStrategy",
+    "TopPStrategy",
+    "SamplingPipeline",
 ]

astrai/inference/cache.py

@@ -0,0 +1,174 @@
+"""Page-based KV cache with page-table-indirected read/write.
+
+Provides:
+  - PagedCache: paged KV cache combining page pool and tensor storage.
+"""
+
+from typing import Dict, List, Tuple
+
+import torch
+from torch import Tensor
+
+STOP = object()
+
+
+def page_hash(token_ids: List[int], page_idx: int, page_size: int) -> int:
+    start = page_idx * page_size
+    end = min(start + page_size, len(token_ids))
+    h = 0
+    for i in range(start, end):
+        h = (h * 31 + token_ids[i]) & 0xFFFFFFFFFFFFFFFF
+    return h
+
+
+class PagedCache:
+    """Paged KV cache with page-table-indirected read/write.
+
+    Combines:
+      - Page pool (ref-counted alloc/free via bitmask)
+      - KV tensor storage (k_cache, v_cache)
+      - Prefix-cache hash lookup (page_content_hash -> physical_page_idx)
+
+    Call :meth:`bind` to obtain a batch view for the attention layers.
+    """
+
+    def __init__(
+        self,
+        n_layers: int,
+        n_pages: int,
+        page_size: int,
+        n_kv_heads: int,
+        head_dim: int,
+        device: torch.device,
+        dtype: torch.dtype,
+    ):
+        self.page_size = page_size
+        self._free_mask = (1 << n_pages) - 1
+        self._refs: List[int] = [0] * n_pages
+        self.k_cache = torch.empty(
+            (n_layers, n_pages, page_size, n_kv_heads, head_dim),
+            device=device,
+            dtype=dtype,
+        )
+        self.v_cache = torch.empty(
+            (n_layers, n_pages, page_size, n_kv_heads, head_dim),
+            device=device,
+            dtype=dtype,
+        )
+        self._page_to_hash: Dict[int, int] = {}
+        self._hash_to_page: Dict[int, int] = {}
+
+    def record_page(
+        self, page_idx: int, token_ids: List[int], logical_page_idx: int
+    ) -> None:
+        h = page_hash(token_ids, logical_page_idx, self.page_size)
+        old_h = self._page_to_hash.pop(page_idx, None)
+        if old_h is not None:
+            self._hash_to_page.pop(old_h, None)
+        self._page_to_hash[page_idx] = h
+        self._hash_to_page[h] = page_idx
+
+    def lookup_prefix(self, token_ids: List[int]) -> List[int]:
+        full_pages = len(token_ids) // self.page_size
+        hits: List[int] = []
+        for i in range(full_pages):
+            h = page_hash(token_ids, i, self.page_size)
+            p = self._hash_to_page.get(h)
+            if p is None:
+                break
+            hits.append(p)
+        return hits
+
+    def inc_ref(self, idx: int) -> None:
+        self._refs[idx] += 1
+
+    def alloc(self) -> int:
+        lsb = self._free_mask & -self._free_mask
+        if lsb == 0:
+            return -1
+        idx = lsb.bit_length() - 1
+        self._free_mask ^= lsb
+        self._refs[idx] = 1
+        return idx
+
+    def alloc_n(self, n: int) -> List[int]:
+        pages = [self.alloc() for _ in range(n)]
+        if any(p < 0 for p in pages):
+            for p in pages:
+                if p >= 0:
+                    self.free(p)
+            return []
+        return pages
+
+    def free(self, idx: int) -> None:
+        self._refs[idx] -= 1
+        if self._refs[idx] == 0:
+            self._free_mask |= 1 << idx
+            h = self._page_to_hash.pop(idx, None)
+            if h is not None:
+                self._hash_to_page.pop(h, None)
+
+    def bind(self, page_table: Tensor, total_len: int = 0) -> "CacheView":
+        return CacheView(self, page_table, total_len)
+
+    def write(
+        self, layer_id: int, page_table: Tensor, start_pos: int, k: Tensor, v: Tensor
+    ) -> None:
+        seq_len = k.size(1)
+        if seq_len == 0:
+            return
+        page_size = self.page_size
+        written = 0
+        first_page = start_pos // page_size
+        last_page = (start_pos + seq_len - 1) // page_size
+        for pi in range(first_page, last_page + 1):
+            phys_pages = page_table[:, pi]
+            page_start = pi * page_size
+            write_start = max(page_start, start_pos)
+            write_end = min(page_start + page_size, start_pos + seq_len)
+            offset = write_start - page_start
+            chunk = write_end - write_start
+            self.k_cache[layer_id, phys_pages, offset : offset + chunk] = k[
+                :, written : written + chunk
+            ]
+            self.v_cache[layer_id, phys_pages, offset : offset + chunk] = v[
+                :, written : written + chunk
+            ]
+            written += chunk
+
+    def gather(self, layer_id: int, page_table: Tensor) -> Tuple[Tensor, Tensor]:
+        k_parts, v_parts = [], []
+        for pi in range(page_table.size(1)):
+            phys_pages = page_table[:, pi]
+            if not (phys_pages >= 0).any():
+                break
+            k_parts.append(self.k_cache[layer_id, phys_pages])
+            v_parts.append(self.v_cache[layer_id, phys_pages])
+        k = torch.cat(k_parts, dim=1)
+        v = torch.cat(v_parts, dim=1)
+        return k, v
+
+
+class CacheView:
+    """Per-batch view that bundles PagedCache + page_table + total_len.
+
+    Attention layers receive this as ``paged_cache`` and only see
+    ``write()`` / ``gather()``, never raw page tables or length params.
+    """
+
+    __slots__ = ("_cache", "_page_table", "_total_len")
+
+    def __init__(self, cache: PagedCache, page_table: Tensor, total_len: int = 0):
+        self._cache = cache
+        self._page_table = page_table
+        self._total_len = total_len
+
+    def write(self, layer_id: int, start_pos: int, k: Tensor, v: Tensor) -> None:
+        self._cache.write(layer_id, self._page_table, start_pos, k, v)
+
+    def gather(self, layer_id: int) -> Tuple[Tensor, Tensor]:
+        k, v = self._cache.gather(layer_id, self._page_table)
+        if self._total_len:
+            k = k[:, : self._total_len]
+            v = v[:, : self._total_len]
+        return k, v

astrai/inference/engine.py

@@ -1,21 +1,42 @@
-"""Unified inference engine."""
+"""Unified inference engine for continuous batching.
 
+Layers:
+  - GenerationParams:    Immutable value object for sampling parameters.
+  - GenerationRequest:   User-facing request DTO with validation.
+  - _Result:             Thread-safe token accumulator (Observer pattern).
+  - InferenceEngine:     Facade over InferenceScheduler + async wrapper.
+"""
+
+import asyncio
 import gc
-import logging
 import threading
-from typing import Any, Dict, Generator, List, Optional, Union
+from dataclasses import dataclass
+from typing import Any, AsyncGenerator, Dict, Generator, List, Optional, Union
 
 import torch
 import torch.nn as nn
 
+from astrai.inference.cache import STOP
 from astrai.inference.scheduler import InferenceScheduler
 from astrai.tokenize import AutoTokenizer
 
-logger = logging.getLogger(__name__)
+
+@dataclass(frozen=True)
+class GenerationParams:
+    """Immutable value object for sampling hyperparameters."""
+
+    top_k: int = 50
+    top_p: float = 1.0
+    temperature: float = 1.0
+    max_tokens: int = 1024
 
 
 class GenerationRequest:
-    """Request parameters for text generation."""
+    """Request parameters for text generation.
+
+    Encapsulates messages, sampling parameters (via GenerationParams),
+    and streaming preference for a single generation request.
+    """
 
     def __init__(
         self,
@@ -26,17 +47,44 @@ class GenerationRequest:
         max_len: int = 1024,
         stream: bool = False,
     ):
-        self.messages = messages
-        self.top_k = top_k
-        self.top_p = top_p
-        self.temperature = temperature
-        self.max_len = max_len
-        self.stream = stream
+        """Initializes a generation request.
 
+        Args:
+            messages: Conversation history as list of {"role": ..., "content": ...}.
+            top_k: Top-k sampling count (0 disables).
+            top_p: Nucleus sampling probability threshold.
+            temperature: Sampling temperature.
+            max_len: Maximum tokens to generate.
+            stream: Whether to return output as a token stream.
+        """
+        self.messages = messages
+        self.params = GenerationParams(
+            top_k=top_k,
+            top_p=top_p,
+            temperature=temperature,
+            max_tokens=max_len,
+        )
+        self.stream = stream
         self._validate()
 
+    @property
+    def top_k(self) -> int:
+        return self.params.top_k
+
+    @property
+    def top_p(self) -> float:
+        return self.params.top_p
+
+    @property
+    def temperature(self) -> float:
+        return self.params.temperature
+
+    @property
+    def max_len(self) -> int:
+        return self.params.max_tokens
+
     def _validate(self):
-        """Validate request parameters."""
+        """Validates sampling parameter ranges."""
         if not (isinstance(self.top_k, int) and self.top_k >= 0):
             raise ValueError("top_k must be a non-negative integer")
         if not (0.0 <= self.top_p <= 1.0):
@@ -46,50 +94,101 @@ class GenerationRequest:
 
 
 class _Result:
-    """Unified result holder for streaming/non-streaming modes."""
+    """Thread-safe token accumulator for streaming and non-streaming modes.
 
-    def __init__(self, count: int = 1, stream: bool = False):
-        self._stream = stream
-        self._lock = threading.Lock()
+    Supports multiple concurrent generation tasks with per-index result tracking.
+    Uses a threading.Condition for efficient completion notification
+    and a threading.Event for streaming wakeup.
+    """
+
+    def __init__(self, count: int = 1):
+        """Initializes the accumulator.
+
+        Args:
+            count: Number of concurrent generation tasks to track.
+        """
+        self._cond = threading.Condition()
         self._event = threading.Event()
         self.tokens: List[str] = []
-        self.results: List[str] = [""] * count if count > 1 else [""]
-        self.done_flags: List[bool] = [False] * count
-        self._completed_count = 0
+        self.results: List[str] = [""] * count
+        self._done: List[bool] = [False] * count
+        self._completed = 0
+        self._total = count
 
     def append(self, token: str, idx: int = 0):
-        with self._lock:
-            if self._stream:
-                self.tokens.append(token)
+        """Appends a token to the result buffer.
+
+        In non-streaming mode, tokens are concatenated into results[idx].
+        The sentinel STOP marks a task as complete.
+
+        Args:
+            token: The decoded token string, or STOP sentinel.
+            idx: Index of the generation task this token belongs to.
+        """
+        with self._cond:
+            self.tokens.append(token)
+            if token is not STOP:
+                self.results[idx] += token
             else:
-                if token == "[DONE]":
-                    if not self.done_flags[idx]:
-                        self.done_flags[idx] = True
-                        self._completed_count += 1
-                        if self._completed_count == len(self.results):
-                            self._event.set()
-                else:
-                    self.results[idx] += token
-        self._event.set()
+                if not self._done[idx]:
+                    self._done[idx] = True
+                    self._completed += 1
+                    self._cond.notify_all()
+            self._event.set()
 
     def pop_all(self) -> List[str]:
-        with self._lock:
-            tokens = self.tokens.copy()
-            self.tokens.clear()
-            if not tokens:
-                self._event.clear()
-            return tokens
+        """Returns and clears all accumulated tokens.
 
-    def wait(self, timeout: float = None) -> bool:
+        Returns:
+            List of token strings since the last call.
+        """
+        with self._cond:
+            out = self.tokens.copy()
+            self.tokens.clear()
+            if not out:
+                self._event.clear()
+            return out
+
+    def wait(self, timeout: Optional[float] = None) -> bool:
+        """Blocks until new tokens arrive or the timeout expires.
+
+        Args:
+            timeout: Maximum wait time in seconds (None = infinite).
+
+        Returns:
+            True if the event was set (new data available), False on timeout.
+        """
         return self._event.wait(timeout=timeout)
 
+    def wait_completion(self) -> None:
+        """Blocks until all tasks complete (non-streaming).
+
+        Uses a Condition to sleep efficiently instead of busy-waiting.
+        The calling thread is parked until a STOP signal arrives.
+        """
+        with self._cond:
+            self._cond.wait_for(lambda: self._completed >= self._total)
+
     def get_results(self) -> List[str]:
-        with self._lock:
+        """Returns all accumulated results for non-streaming mode.
+
+        Returns:
+            List of complete generated strings, one per task index.
+        """
+        with self._cond:
             return self.results.copy()
 
 
 class InferenceEngine:
-    """Unified inference engine for continuous batching."""
+    """Unified inference engine backed by continuous-batching scheduler.
+
+    Usage:
+        with InferenceEngine(model, tokenizer) as engine:
+            for token in engine.generate("hello", stream=True):
+                print(token, end="")
+
+            text = engine.generate("hello")
+    """
 
     def __init__(
         self,
@@ -97,55 +196,37 @@ class InferenceEngine:
         tokenizer: AutoTokenizer,
         max_batch_size: int = 1,
         max_seq_len: Optional[int] = None,
-        max_prefix_len: int = 512,
-        cache_capacity: int = 1000,
+        max_prompt_len: int = 2048,
+        page_size: int = 128,
     ):
-        """
-        Initialize inference engine with separate model and tokenizer.
+        """Initializes the inference engine.
 
         Args:
-            model: The language model for inference (nn.Module, e.g., Transformer)
-            tokenizer: The tokenizer for encoding/decoding text
-            config: Model configuration
-            max_batch_size: Maximum batch size for continuous batching
-            max_seq_len: Maximum sequence length (defaults to config.max_len)
-            max_prefix_len: Maximum prefix length for cache (default: 512)
-            cache_capacity: Maximum number of cached prefixes (default: 1000)
+            model: The model instance.
+            tokenizer: The tokenizer instance.
+            max_batch_size: Maximum number of concurrent tasks.
+            max_seq_len: Maximum sequence length.
+            max_prompt_len: Maximum prompt tokens.
+            compile: Whether to compile the model with torch.compile.
+            page_size: Number of tokens per KV cache page.
         """
         self.model = model
         self.tokenizer = tokenizer
-
-        # Get device and dtype from model parameters
-        try:
-            first_param = next(model.parameters())
-            device = first_param.device
-            dtype = first_param.dtype
-        except StopIteration:
-            # Model has no parameters, use default device/dtype
-            device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-            dtype = torch.float32
-
         self.scheduler = InferenceScheduler(
             model=self.model,
             tokenizer=self.tokenizer,
             max_batch_size=max_batch_size,
             max_seq_len=max_seq_len,
-            max_prefix_len=max_prefix_len,
-            cache_capacity=cache_capacity,
-            device=device,
-            dtype=dtype,
+            max_prompt_len=max_prompt_len,
+            page_size=page_size,
         )
 
-        self.kv_cache = self.scheduler.kv_cache
-        self.seq_mask = self.scheduler.seq_mask
-
         self.scheduler.start()
 
     def __enter__(self):
         return self
 
     def __exit__(self, exc_type, exc_val, exc_tb):
-        """Handle exceptions on exit."""
         self.shutdown()
         return False
 
@@ -157,46 +238,106 @@ class InferenceEngine:
         temperature: float = 1.0,
         top_p: float = 1.0,
         top_k: int = 50,
-        abort_on_exception: bool = True,
     ) -> Union[Generator[str, None, None], str, List[str]]:
-        """Unified generation interface.
+        """Generates text from a prompt.
 
         Args:
-            abort_on_exception: If True, abort the generation when consumer
-                stops iterating (GeneratorExit/StopIteration). Default: True.
+            prompt: Single string or list of strings for batch generation.
+            stream: If True, returns a generator yielding tokens one by one.
+            max_tokens: Maximum number of tokens to generate.
+            temperature: Sampling temperature.
+            top_p: Nucleus sampling probability threshold.
+            top_k: Top-k sampling count (0 disables).
+
+        Returns:
+            Generator (stream=True), single string (non-stream, single prompt),
+            or list of strings (non-stream, batch prompts).
         """
         is_batch = isinstance(prompt, list)
         prompts = prompt if is_batch else [prompt]
 
         if stream:
             return self._generate_streaming(
-                prompts,
-                is_batch,
-                max_tokens,
-                temperature,
-                top_p,
-                top_k,
-                abort_on_exception,
+                prompts, is_batch, max_tokens, temperature, top_p, top_k
             )
         else:
             return self._generate_non_streaming(
                 prompts, is_batch, max_tokens, temperature, top_p, top_k
             )
 
+    def generate_async(
+        self,
+        prompt: str,
+        max_tokens: int = 1024,
+        temperature: float = 1.0,
+        top_p: float = 1.0,
+        top_k: int = 50,
+    ) -> AsyncGenerator[str, None]:
+        """Async streaming generator that does not block the event loop.
+
+        Runs the synchronous generator in a background thread pool executor,
+        yielding tokens to the async consumer as they arrive.
+
+        Args:
+            prompt: Input text to generate from.
+            max_tokens: Maximum tokens to generate.
+            temperature: Sampling temperature.
+            top_p: Nucleus sampling threshold.
+            top_k: Top-k sampling count.
+
+        Yields:
+            Decoded token strings as they are generated.
+        """
+        sync_gen = self._generate_streaming(
+            [prompt], False, max_tokens, temperature, top_p, top_k
+        )
+
+        async def _agen():
+            loop = asyncio.get_event_loop()
+            while True:
+                token = await loop.run_in_executor(None, self._next_token, sync_gen)
+                if token is None:
+                    break
+                yield token
+
+        return _agen()
+
+    @staticmethod
+    def _next_token(gen: Generator) -> Optional[str]:
+        """Retrieves the next token from a synchronous generator.
+
+        Args:
+            gen: A synchronous generator yielding token strings.
+
+        Returns:
+            The next token, or None if the generator is exhausted.
+        """
+        try:
+            return next(gen)
+        except StopIteration:
+            return None
+
     def generate_with_request(
         self, request: GenerationRequest
     ) -> Union[Generator[str, None, None], str, List[str]]:
-        """Generate with GenerationRequest object."""
-        # Use tokenizer's chat template with messages
-        prompt = self.tokenizer.apply_chat_template(request.messages, tokenize=False)
+        """Generates text from a structured GenerationRequest.
 
+        Applies the chat template to the request's messages before generation.
+
+        Args:
+            request: A GenerationRequest with messages and parameters.
+
+        Returns:
+            Generator, string, or list of strings (see generate()).
+        """
+        prompt = self.tokenizer.apply_chat_template(request.messages, tokenize=False)
         return self.generate(
             prompt=prompt,
             stream=request.stream,
-            max_tokens=request.max_len,
-            temperature=request.temperature,
-            top_p=request.top_p,
-            top_k=request.top_k,
+            max_tokens=request.params.max_tokens,
+            temperature=request.params.temperature,
+            top_p=request.params.top_p,
+            top_k=request.params.top_k,
         )
 
     def _generate_streaming(
@@ -207,18 +348,27 @@ class InferenceEngine:
         temperature: float,
         top_p: float,
         top_k: int,
-        abort_on_exception: bool = True,
-    ) -> Union[Generator[str, None, None], List[Generator[str, None, None]]]:
-        """Generate with streaming output.
+    ) -> Generator[str, None, None]:
+        """Internal streaming generator.
+
+        Polls the _Result accumulator in a loop, yielding tokens as they arrive.
+        Cleans up the scheduler task on GeneratorExit.
 
         Args:
-            abort_on_exception: If True, abort the task when generator is
-                stopped early by consumer (GeneratorExit/StopIteration).
+            prompts: List of prompts (only first is used; batch not yet supported).
+            is_batch: If True, raises NotImplementedError.
+            max_tokens: Maximum tokens to generate.
+            temperature: Sampling temperature.
+            top_p: Nucleus sampling threshold.
+            top_k: Top-k sampling count.
+
+        Yields:
+            Decoded token strings.
         """
         if is_batch:
-            raise NotImplementedError("Batch streaming is not implemented yet")
+            raise NotImplementedError("Batch streaming not yet supported")
 
-        result = _Result(stream=True)
+        result = _Result()
 
         task_id = self.scheduler.add_task(
             prompt=prompts[0],
@@ -226,7 +376,7 @@ class InferenceEngine:
             temperature=temperature,
             top_p=top_p,
             top_k=top_k,
-            stream_callback=result.append,
+            stream_callback=lambda tok: result.append(tok, 0),
         )
 
         def gen():
@@ -234,17 +384,14 @@ class InferenceEngine:
                 while True:
                     tokens = result.pop_all()
                     for token in tokens:
-                        if token == "[DONE]":
+                        if token is STOP:
                             return
                         yield token
-                    result.wait(timeout=0.05)
-            except Exception:
-                # Consumer stopped iterating - abort the task
-                if abort_on_exception:
-                    self.scheduler.remove_task(task_id)
-                raise
+                    if not result.wait(timeout=0.05):
+                        pass
+            finally:
+                self.scheduler.remove_task(task_id)
 
-        gen.task_id = task_id
         return gen()
 
     def _generate_non_streaming(
@@ -256,36 +403,57 @@ class InferenceEngine:
         top_p: float,
         top_k: int,
     ) -> Union[str, List[str]]:
-        """Generate without streaming."""
+        """Internal non-streaming generator.
+
+        Submits all prompts to the scheduler and waits for all to complete.
+
+        Args:
+            prompts: List of prompt strings.
+            is_batch: Whether multiple prompts were provided.
+            max_tokens: Maximum tokens to generate.
+            temperature: Sampling temperature.
+            top_p: Nucleus sampling threshold.
+            top_k: Top-k sampling count.
+
+        Returns:
+            Single string for one prompt, list of strings for batch.
+        """
         result = _Result(count=len(prompts))
+        task_ids = []
 
         for i, p in enumerate(prompts):
-            # Create closure to capture current index value using factory function
-            def make_callback(idx):
-                def callback(token):
-                    result.append(idx, token)
 
-                return callback
+            def make_cb(idx):
+                return lambda tok: result.append(tok, idx)
 
-            self.scheduler.add_task(
+            task_id = self.scheduler.add_task(
                 prompt=p,
                 max_tokens=max_tokens,
                 temperature=temperature,
                 top_p=top_p,
                 top_k=top_k,
-                stream_callback=make_callback(i),
+                stream_callback=make_cb(i),
             )
+            task_ids.append(task_id)
 
-        result.wait()
-        results = result.get_results()
-        return results if is_batch else results[0]
+        result.wait_completion()
+
+        for task_id in task_ids:
+            self.scheduler.remove_task(task_id)
+
+        res = result.get_results()
+        return res if is_batch else res[0]
 
     def get_stats(self) -> Dict[str, Any]:
-        """Get engine statistics."""
+        """Returns current engine statistics.
+
+        Returns:
+            Dict with total_tasks, total_tokens, active_tasks, waiting_queue.
+        """
         return self.scheduler.get_stats()
 
     def shutdown(self) -> None:
-        """Shutdown the engine and release all resources."""
+        """Shuts down the engine, stops the scheduler, and frees GPU memory."""
         self.scheduler.stop()
         if torch.cuda.is_available():
             torch.cuda.empty_cache()

astrai/inference/sampling.py

@@ -0,0 +1,178 @@
+"""Composable sampling strategies for logit transformation.
+
+Implements the Strategy pattern: each sampling technique
+(temperature, top-k, top-p) is a pluggable strategy that
+can be composed into a pipeline.
+
+All strategies accept both scalar and per-sample tensor
+parameters, so a single pipeline works for any batch size.
+"""
+
+from abc import ABC, abstractmethod
+from typing import List, Union
+
+import torch
+from torch import Tensor
+
+
+class BaseSamplingStrategy(ABC):
+    """Abstract base for a logit transformation strategy."""
+
+    @abstractmethod
+    def apply(self, logits: Tensor, filter_value: float = -float("inf")) -> Tensor:
+        """Applies the strategy to logits.
+
+        Args:
+            logits: Raw logits tensor (batch, vocab_size).
+            filter_value: Value assigned to filtered-out positions.
+
+        Returns:
+            Transformed logits tensor.
+        """
+
+
+class TemperatureStrategy(BaseSamplingStrategy):
+    """Divides logits by temperature to control randomness.
+
+    Args:
+        temperature: Scalar or ``[batch]`` tensor.
+    """
+
+    def __init__(self, temperature: Union[float, Tensor] = 1.0):
+        self.temperature = temperature
+
+    def apply(self, logits, filter_value=-float("inf")):
+        t = self.temperature
+        if isinstance(t, Tensor):
+            if (t != 1.0).any():
+                logits = logits / t.to(logits.device, non_blocking=True).view(-1, 1)
+        elif t != 1.0:
+            logits = logits / t
+        return logits
+
+
+class TopKStrategy(BaseSamplingStrategy):
+    """Keeps only the top-k logits, setting the rest to filter_value.
+
+    Args:
+        top_k: Scalar or ``[batch]`` tensor (0 disables).
+    """
+
+    def __init__(self, top_k: Union[int, Tensor] = 0):
+        self.top_k = top_k
+
+    def apply(self, logits, filter_value=-float("inf")):
+        tk = self.top_k
+        if isinstance(tk, Tensor):
+            max_k = int(tk.max().item())
+            if max_k <= 0:
+                return logits
+            k = min(max_k, logits.size(-1))
+        elif tk > 0:
+            k = min(tk, logits.size(-1))
+        else:
+            return logits
+        thresholds = torch.topk(logits, k, dim=-1)[0][..., -1:]
+        logits[logits < thresholds] = filter_value
+        return logits
+
+
+class TopPStrategy(BaseSamplingStrategy):
+    """Nucleus (top-p) filtering: keeps the smallest set of tokens whose
+    cumulative probability exceeds top_p.
+
+    Args:
+        top_p: Scalar or ``[batch]`` tensor (1.0 disables).
+    """
+
+    def __init__(self, top_p: Union[float, Tensor] = 1.0):
+        self.top_p = top_p
+
+    def _apply(self, logits, top_p, filter_value):
+        sorted_logits, sorted_indices = torch.sort(logits, descending=True, dim=-1)
+        cum_probs = torch.cumsum(torch.softmax(sorted_logits, dim=-1), dim=-1)
+        remove = cum_probs > top_p
+        remove[..., 1:] = remove[..., :-1].clone()
+        remove[..., 0] = False
+        mask = torch.zeros_like(logits, dtype=torch.bool)
+        mask.scatter_(1, sorted_indices, remove)
+        logits[mask] = filter_value
+        return logits
+
+    def apply(self, logits, filter_value=-float("inf")):
+        tp = self.top_p
+        if isinstance(tp, Tensor):
+            tp = tp.to(logits.device, non_blocking=True)
+            if (tp < 1.0).any():
+                logits = self._apply(logits, tp.view(-1, 1), filter_value)
+        elif tp < 1.0:
+            logits = self._apply(logits, tp, filter_value)
+        return logits
+
+
+class SamplingPipeline(BaseSamplingStrategy):
+    """Composes multiple sampling strategies into a single transformation.
+
+    Strategies are applied sequentially in the order they are provided,
+    matching the original temperature -> top-k -> top-p ordering.
+
+    Usage::
+
+        pipeline = SamplingPipeline([
+            TemperatureStrategy(0.8),
+            TopKStrategy(50),
+            TopPStrategy(0.95),
+        ])
+        logits = pipeline.apply(logits)
+        token = pipeline.sample(logits)       # softmax + multinomial
+    """
+
+    def __init__(self, strategies: List[BaseSamplingStrategy]):
+        self.strategies = strategies
+
+    def apply(self, logits, filter_value=-float("inf")):
+        for strategy in self.strategies:
+            logits = strategy.apply(logits, filter_value)
+        return logits
+
+    @torch.no_grad()
+    def sample(self, logits: Tensor, filter_value: float = -float("inf")) -> Tensor:
+        """Apply strategies then sample (softmax + multinomial).
+
+        Args:
+            logits: Raw logits ``[batch, vocab_size]``.
+
+        Returns:
+            Sampled token IDs ``[batch]``.
+        """
+        return torch.multinomial(
+            torch.softmax(self.apply(logits, filter_value), dim=-1),
+            num_samples=1,
+        ).squeeze(-1)
+
+
+@torch.inference_mode()
+def sample(
+    logits: Tensor,
+    temperature: Union[float, Tensor] = 1.0,
+    top_k: Union[int, Tensor] = 0,
+    top_p: Union[float, Tensor] = 1.0,
+    filter_value: float = -float("inf"),
+) -> Tensor:
+    """Apply sampling strategies then sample (softmax + multinomial).
+
+    Shortcut for ``SamplingPipeline(...).sample(logits)``.
+
+    Args:
+        logits: Raw logits ``[batch, vocab_size]``.
+
+    Returns:
+        Sampled token IDs ``[batch]``.
+    """
+    return SamplingPipeline(
+        [
+            TemperatureStrategy(temperature),
+            TopKStrategy(top_k),
+            TopPStrategy(top_p),
+        ]
+    ).sample(logits, filter_value)

astrai/inference/scheduler.py

@@ -1,148 +1,25 @@
-"""Inference scheduler for continuous batching."""
+"""Inference scheduler for single-GPU continuous batching with paged KV cache."""
 
+import logging
 import threading
 import time
 import uuid
+from enum import Enum
 from typing import Any, Callable, Dict, List, Optional, Tuple
 
 import torch
 from torch import Tensor
 
+from astrai.inference.cache import STOP, PagedCache
+from astrai.inference.sampling import sample
 from astrai.model.automodel import AutoModel
-from astrai.tokenize import AutoTokenizer
+from astrai.tokenize.tokenizer import AutoTokenizer
+
+logger = logging.getLogger(__name__)
 
 
-class RadixNode:
-    """Radix tree node for prefix cache."""
-
-    def __init__(self):
-        self.children: Dict[int, "RadixNode"] = {}  # token_id -> child node
-        self.hash: Optional[int] = None  # 64-bit hash of the prefix
-        self.slot: int = -1  # KV Cache slot, valid only for leaf nodes
-        self.ref_count: int = 0  # number of tasks referencing this prefix
-        self.last_access: float = 0.0  # timestamp for LRU
-        self.token_sequence: list = []  # full token sequence from root to this node
-
-
-class PrefixCacheManager:
-    """Prefix cache manager using Radix tree with LRU eviction."""
-
-    def __init__(self, max_capacity: int = 1000, base: int = 131, mod: int = 10**9 + 7):
-        self.root = RadixNode()
-        self.base = base
-        self.mod = mod
-        self.max_capacity = max_capacity
-        self.lru: List[Tuple[float, RadixNode]] = []  # (timestamp, node) for LRU
-
-    def insert(self, token_ids: Tuple[int, ...], slot: int) -> None:
-        """Insert a prefix, increase ref_count if already exists, otherwise create new node."""
-        node = self.root
-        path = []
-        h = 0
-        for i, token_id in enumerate(token_ids):
-            if token_id not in node.children:
-                node.children[token_id] = RadixNode()
-            node = node.children[token_id]
-            h = (h * self.base + token_id) % self.mod
-            node.hash = h
-            path.append(token_id)
-            node.token_sequence = list(
-                path
-            )  # store full sequence for exact verification
-
-        # Leaf node: set slot and increase ref_count
-        if node.slot == -1:
-            node.slot = slot
-        node.ref_count += 1
-        node.last_access = time.time()
-        self._update_lru(node)
-        self._evict_if_needed()
-
-    def find_longest_prefix(self, token_ids: List[int]) -> Optional[Tuple[int, int]]:
-        """Find longest matching prefix, return (prefix_len, slot).
-
-        During traversal, compute hash per token and compare with node hash.
-        If hash matches, perform full token sequence verification to avoid
-        hash collision errors.
-        """
-        node = self.root
-        best_len = 0
-        best_slot = -1
-        h = 0
-
-        for i, token_id in enumerate(token_ids):
-            if token_id not in node.children:
-                break
-            node = node.children[token_id]
-            h = (h * self.base + token_id) % self.mod
-            if node.hash == h:  # hash matches
-                # Exact verification: compare full token sequence
-                if node.token_sequence == token_ids[: i + 1]:
-                    best_len = i + 1
-                    best_slot = node.slot
-                    node.last_access = time.time()
-                    self._update_lru(node)
-
-        if best_len > 0:
-            return (best_len, best_slot)
-        return None
-
-    def release(self, token_ids: Tuple[int, ...]) -> None:
-        """Release reference to a prefix, decrease ref_count. If zero, mark as evictable."""
-        node = self.root
-        for token_id in token_ids:
-            if token_id not in node.children:
-                return
-            node = node.children[token_id]
-        if node.ref_count > 0:
-            node.ref_count -= 1
-            if node.ref_count == 0:
-                node.slot = -1  # slot can be reused
-
-    def _update_lru(self, node: RadixNode) -> None:
-        """Update LRU list, move node to most recently used position."""
-        self.lru = [(ts, n) for (ts, n) in self.lru if n is not node]
-        self.lru.append((node.last_access, node))
-
-    def _evict_if_needed(self) -> None:
-        """If cache entries exceed capacity, evict least recently used leaf nodes (ref_count must be 0)."""
-        if len(self.lru) <= self.max_capacity:
-            return
-        # Sort by timestamp
-        self.lru.sort(key=lambda x: x[0])
-        for ts, node in self.lru:
-            if node.ref_count == 0:
-                # Remove leaf node from tree (need to recursively delete empty branches)
-                self._remove_node(node)
-                self.lru.remove((ts, node))
-                if len(self.lru) <= self.max_capacity:
-                    break
-
-    def _remove_node(
-        self,
-        node: RadixNode,
-        parent: Optional[RadixNode] = None,
-        child_key: Optional[int] = None,
-    ) -> None:
-        """Remove node from tree, including empty parent nodes."""
-        # First, recursively remove all children
-        for child_key, child_node in list(node.children.items()):
-            self._remove_node(child_node, node, child_key)
-
-        # Clear the node's leaf properties
-        node.slot = -1
-        node.hash = None
-        node.token_sequence = []
-        node.children.clear()
-
-        # If this node has no children and has a parent, remove the reference from parent
-        if parent is not None and child_key is not None and len(node.children) == 0:
-            if child_key in parent.children:
-                del parent.children[child_key]
-
-
-class TaskStatus:
-    """Task state for continuous batching."""
+class TaskStatus(Enum):
+    """Task states in the continuous batching lifecycle."""
 
     PENDING = "pending"
     RUNNING = "running"
@@ -151,7 +28,7 @@ class TaskStatus:
 
 
 class Task:
-    """Individual task for continuous batching."""
+    """Represents a single generation request with paged KV cache tracking."""
 
     def __init__(
         self,
@@ -174,60 +51,35 @@ class Task:
         self.output_ids: List[int] = []
         self.input_tokens: int = 0
         self.output_tokens: int = 0
-        self.slot: int = -1
-        self.prefix_len: int = 0  # prefix cache matched length
+        self.page_table: List[int] = []
+        self.n_pages: int = 0
+        self._prefix_cached_tokens: int = 0
         self.arrival_time = time.time()
         self.finish_time: Optional[float] = None
-
         self.stream_callback = stream_callback
+        self._pages_freed: bool = False
+
+    @property
+    def next_pos(self) -> int:
+        return self.input_tokens + len(self.output_ids)
 
     def is_finished(self, stop_ids: List[int]) -> bool:
-        """Check if task is finished."""
-        return (
-            bool(self.output_ids and self.output_ids[-1] in stop_ids)
-            or self.output_tokens >= self.max_tokens
-        )
-
-
-def apply_sampling_strategies(
-    logits: Tensor,
-    temperature: float,
-    top_k: int,
-    top_p: float,
-    filter_value: float = -float("inf"),
-) -> Tensor:
-    """Apply sampling strategies to the logits tensor."""
-    # Clone logits to avoid inplace updates on inference tensor
-    logits = logits.clone()
-
-    if temperature != 1.0:
-        logits = logits / temperature
-
-    if top_k > 0:
-        top_k = min(top_k, logits.size(-1))
-        indices_to_remove = logits < torch.topk(logits, top_k, dim=-1)[0][..., -1, None]
-        logits[indices_to_remove] = filter_value
-
-    if top_p < 1.0:
-        sorted_logits, sorted_indices = torch.sort(logits, descending=True, dim=-1)
-        cumulative_probs = torch.cumsum(torch.softmax(sorted_logits, dim=-1), dim=-1)
-
-        sorted_indices_to_remove = cumulative_probs > top_p
-        sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
-        sorted_indices_to_remove[..., 0] = 0
-
-        indices_to_remove = torch.zeros_like(logits, dtype=torch.bool)
-        indices_to_remove.scatter_(
-            dim=1, index=sorted_indices, src=sorted_indices_to_remove
-        )
-
-        logits[indices_to_remove] = filter_value
-
-    return logits
+        if self.output_tokens >= self.max_tokens:
+            return True
+        if self.output_ids and self.output_ids[-1] in stop_ids:
+            return True
+        return False
 
 
 class InferenceScheduler:
-    """Inference scheduler with continuous batching support."""
+    """Continuous batching scheduler with paged KV cache.
+
+    Runs a background generation loop with four phases per iteration:
+      1. Cleanup finished tasks and release resources.
+      2. Refill active batch from the waiting queue.
+      3. Prefill newly activated tasks.
+      4. Decode the largest same-position group of active tasks.
+    """
 
     def __init__(
         self,
@@ -235,10 +87,10 @@ class InferenceScheduler:
         tokenizer: AutoTokenizer,
         max_batch_size: int = 16,
         max_seq_len: Optional[int] = None,
-        max_prefix_len: int = 512,
-        cache_capacity: int = 1000,
-        device: str = "cuda",
-        dtype: torch.dtype = torch.bfloat16,
+        max_prompt_len: int = 512,
+        page_size: int = 64,
+        device: Optional[str] = None,
+        dtype: Optional[torch.dtype] = None,
     ):
         config = model.config
 
@@ -246,42 +98,26 @@ class InferenceScheduler:
         self.tokenizer = tokenizer
         self.max_batch_size = max_batch_size
         self.max_seq_len = max_seq_len or config.max_len
-        self.max_prefix_len = max_prefix_len
+        self.max_prompt_len = max_prompt_len
+        self.page_size = page_size
         self.device = device or next(model.parameters()).device
         self.dtype = dtype or next(model.parameters()).dtype
 
-        # Initialize prefix cache
-        self.prefix_cache = PrefixCacheManager(max_capacity=cache_capacity)
-
-        num_kv_heads = config.n_kv_heads
+        n_kv_heads = config.n_kv_heads
         head_dim = config.dim // config.n_heads
         n_layers = config.n_layers
+        n_pages = (
+            max_batch_size * (self.max_seq_len + page_size) + page_size - 1
+        ) // page_size
 
-        k_cache = torch.empty(
-            (
-                max_batch_size,
-                self.max_seq_len,
-                n_layers,
-                num_kv_heads,
-                head_dim,
-            ),
-            device=self.device,
-            dtype=self.dtype,
-        )
-        v_cache = torch.empty(
-            (
-                max_batch_size,
-                self.max_seq_len,
-                n_layers,
-                num_kv_heads,
-                head_dim,
-            ),
-            device=self.device,
-            dtype=self.dtype,
-        )
-        self.kv_cache = (k_cache, v_cache)
-        self.seq_mask = torch.ones(
-            (max_batch_size, self.max_seq_len), device=self.device, dtype=torch.bool
+        self.page_cache = PagedCache(
+            n_layers,
+            n_pages,
+            page_size,
+            n_kv_heads,
+            head_dim,
+            self.device,
+            self.dtype,
         )
 
         self.waiting_queue: List[Task] = []
@@ -294,6 +130,9 @@ class InferenceScheduler:
         self._total_tasks = 0
         self._total_tokens = 0
 
+    def _n_pages_for(self, n_tokens: int) -> int:
+        return (n_tokens + self.page_size - 1) // self.page_size
+
     def add_task(
         self,
         prompt: str,
@@ -303,13 +142,10 @@ class InferenceScheduler:
         top_k: int = 50,
         stream_callback: Optional[Callable[[str], None]] = None,
     ) -> str:
-        """Add a new task to the waiting queue."""
         task_id = f"task_{int(time.time())}_{uuid.uuid4().hex[:8]}"
         prompt_ids = self.tokenizer.encode(prompt)
-
-        # Truncate if exceeds max_prefix_len
-        if len(prompt_ids) > self.max_prefix_len:
-            prompt_ids = prompt_ids[: self.max_prefix_len]
+        if len(prompt_ids) > self.max_prompt_len:
+            prompt_ids = prompt_ids[-self.max_prompt_len :]
 
         task = Task(
             task_id=task_id,
@@ -321,16 +157,6 @@ class InferenceScheduler:
             stream_callback=stream_callback,
         )
 
-        # Find longest matching prefix from cache
-        match = self.prefix_cache.find_longest_prefix(prompt_ids)
-        if match:
-            prefix_len, slot = match
-            task.prefix_len = prefix_len
-            task.slot = slot
-        else:
-            task.prefix_len = 0
-            task.slot = -1
-
         with self._lock:
             self.waiting_queue.append(task)
             self._total_tasks += 1
@@ -339,13 +165,28 @@ class InferenceScheduler:
         return task_id
 
     def remove_task(self, task_id: str) -> None:
-        """Remove a task from the scheduler."""
         with self._lock:
+            removed_active = [t for t in self.active_tasks if t.task_id == task_id]
             self.waiting_queue = [t for t in self.waiting_queue if t.task_id != task_id]
             self.active_tasks = [t for t in self.active_tasks if t.task_id != task_id]
 
+        for task in removed_active:
+            if not task._pages_freed:
+                self._free_pages(task.page_table)
+                task.page_table.clear()
+                task.n_pages = 0
+                task._pages_freed = True
+
+    def _free_pages(self, indices: List[int]) -> None:
+        for idx in indices:
+            self.page_cache.free(idx)
+
+    def _record_page_hashes(self, task: Task, start_logical_page: int = 0) -> None:
+        full_pages = len(task.prompt_ids) // self.page_size
+        for i in range(start_logical_page, full_pages):
+            self.page_cache.record_page(task.page_table[i], task.prompt_ids, i)
+
     def _remove_finished_tasks(self) -> None:
-        """Remove finished tasks from active batch."""
         finished = []
         for task in self.active_tasks:
             if task.is_finished(self.tokenizer.stop_ids):
@@ -355,280 +196,213 @@ class InferenceScheduler:
                 self._total_tokens += task.output_tokens
 
         for task in finished:
-            slot = task.slot
-            if slot >= 0 and slot < len(self.active_tasks):
-                self.seq_mask[slot, :] = False
-
-            # Release prefix cache reference
-            if task.prefix_len > 0:
-                self.prefix_cache.release(tuple(task.prompt_ids[: task.prefix_len]))
-
-            task.slot = -1
+            if not task._pages_freed:
+                self._free_pages(task.page_table)
+                task.page_table.clear()
+                task.n_pages = 0
+                task._pages_freed = True
 
         self.active_tasks = [
             t for t in self.active_tasks if t.status != TaskStatus.FINISHED
         ]
 
     def _refill_active_batch(self) -> None:
-        """Refill active batch with waiting tasks."""
-        available_slots = self.max_batch_size - len(self.active_tasks)
-        if available_slots <= 0:
+        available = self.max_batch_size - len(self.active_tasks)
+        if available <= 0:
             return
 
+        to_add: List[Task] = []
         with self._lock:
-            to_add = [
-                self.waiting_queue.pop(0)
-                for _ in range(min(available_slots, len(self.waiting_queue)))
-            ]
-            for task in to_add:
-                task.slot = self._allocate_slot()
-                task.status = TaskStatus.RUNNING
-                self.active_tasks.append(task)
+            n = min(available, len(self.waiting_queue))
+            for _ in range(n):
+                to_add.append(self.waiting_queue.pop(0))
 
-    def _allocate_slot(self) -> int:
-        """Allocate an available slot for a task."""
-        for i in range(self.max_batch_size):
-            if not any(t.slot == i for t in self.active_tasks):
-                return i
-        return -1
+        failed: List[Task] = []
+        for task in to_add:
+            prompt_len = len(task.prompt_ids)
 
-    def _execute_prefill(self, tasks: List[Task]) -> None:
-        """Execute Prefill phase with incremental prefill support."""
-        if not tasks:
-            return
+            hit_pages = self.page_cache.lookup_prefix(task.prompt_ids)
+            cached_tokens = len(hit_pages) * self.page_size
+            for p in hit_pages:
+                self.page_cache.inc_ref(p)
 
-        # Group tasks by prefix cache status
-        fully_cached, partial, full = [], [], []
-        for task in tasks:
-            total_len, prefix_len = len(task.prompt_ids), task.prefix_len
-            if prefix_len == total_len:
-                fully_cached.append(task)
-            elif prefix_len > 0:
-                partial.append(task)
-            else:
-                full.append(task)
+            remaining = prompt_len - cached_tokens
+            n_new = self._n_pages_for(remaining) if remaining > 0 else 0
+            new_pages = self.page_cache.alloc_n(n_new) if n_new > 0 else []
 
-        # Handle fully cached tasks
-        for t in fully_cached:
-            t.input_tokens, t.output_tokens = len(t.prompt_ids), 0
-            if t.slot >= 0:
-                self.seq_mask[t.slot, : t.input_tokens] = True
+            if remaining > 0 and not new_pages:
+                for p in hit_pages:
+                    self.page_cache.free(p)
+                failed.append(task)
+                continue
 
-        if full:
-            self._execute_full_prefill(full)
-        if partial:
-            self._execute_partial_prefill(partial)
+            task.page_table = hit_pages + new_pages
+            task.n_pages = len(task.page_table)
+            task._prefix_cached_tokens = cached_tokens
+            task.status = TaskStatus.RUNNING
+            self.active_tasks.append(task)
 
-    def _execute_full_prefill(self, tasks: List[Task]) -> None:
-        """Execute full prefill for tasks without prefix cache."""
-        if not tasks:
-            return
+        if failed:
+            with self._lock:
+                self.waiting_queue[:0] = failed
 
-        tasks = sorted(tasks, key=lambda t: t.slot)
+    def _execute_prefill(
+        self, tasks: List[Task], prompt_len: int, start_pos: int = 0
+    ) -> None:
+        tasks = sorted(tasks, key=lambda t: t.task_id)
+        batch_sz = len(tasks)
 
-        prompt_lens = [len(task.prompt_ids) for task in tasks]
-        max_len = max(prompt_lens)
+        seq_len = prompt_len - start_pos
+        input_ids = torch.empty(batch_sz, seq_len, dtype=torch.long, device=self.device)
+        input_mask = torch.ones(batch_sz, seq_len, dtype=torch.bool, device=self.device)
 
-        input_ids = torch.zeros(
-            len(tasks), max_len, dtype=torch.long, device=self.device
-        )
-        for i, task in enumerate(tasks):
-            if len(task.prompt_ids) > 0:
-                input_ids[i, : len(task.prompt_ids)] = torch.tensor(
-                    task.prompt_ids, device=self.device
-                )
-
-        if self.tokenizer.pad_id is not None:
-            input_mask = torch.ne(input_ids, self.tokenizer.pad_id)
-        else:
-            input_mask = torch.ones(
-                input_ids.shape, dtype=torch.bool, device=self.device
+        for i, t in enumerate(tasks):
+            input_ids[i] = torch.tensor(
+                t.prompt_ids[start_pos:prompt_len], device=self.device
             )
 
+        page_tables = self._make_page_table_tensor(tasks)
+
         with torch.inference_mode():
             self.model(
                 input_ids,
                 input_mask=input_mask,
-                start_pos=0,
-                persistent_key_values=self.kv_cache,
+                start_pos=start_pos,
+                paged_cache=self.page_cache.bind(page_tables, total_len=prompt_len),
             )
 
-        for i, task in enumerate(tasks):
-            task.input_tokens = prompt_lens[i]
-            task.output_tokens = 0
-            # Insert new prefix into cache
-            self.prefix_cache.insert(tuple(task.prompt_ids), task.slot)
-
-        for task in tasks:
-            if task.slot >= 0:
-                self.seq_mask[task.slot, : task.input_tokens] = True
-
-    def _execute_partial_prefill(self, tasks: List[Task]) -> None:
-        """Execute incremental prefill for tasks with partial prefix cache match."""
-        for task in tasks:
-            total_len = len(task.prompt_ids)
-            prefix_len = task.prefix_len
-
-            if prefix_len >= total_len:
-                task.input_tokens = total_len
-                task.output_tokens = 0
-                continue
-
-            # Get new tokens that need prefill
-            new_ids = task.prompt_ids[prefix_len:]
-            new_len = len(new_ids)
-
-            if new_len == 0:
-                task.input_tokens = total_len
-                task.output_tokens = 0
-                continue
-
-            # Build input for incremental prefill
-            input_ids = torch.tensor([new_ids], dtype=torch.long, device=self.device)
-
-            # Input mask should cover from position 0 to prefix_len + new_len
-            # The prefix part uses cached KV, new part needs computation
-            input_mask = torch.ones(
-                (1, prefix_len + new_len), dtype=torch.bool, device=self.device
-            )
-
-            with torch.inference_mode():
-                self.model(
-                    input_ids,
-                    input_mask=input_mask,
-                    start_pos=prefix_len,
-                    persistent_key_values=self.kv_cache,
-                )
-
-            task.input_tokens = total_len
-            task.output_tokens = 0
-
-            # Insert full prefix into cache (ref_count already increased in add_task)
-            self.prefix_cache.insert(tuple(task.prompt_ids), task.slot)
-
-            if task.slot >= 0:
-                self.seq_mask[task.slot, : task.input_tokens] = True
+        start_logical_page = start_pos // self.page_size
+        for t in tasks:
+            self._record_page_hashes(t, start_logical_page=start_logical_page)
 
     def _execute_decode(self, tasks: List[Task], start_pos: int) -> None:
-        """Execute Decode phase."""
         if not tasks:
             return
 
-        tasks = sorted(tasks, key=lambda t: t.slot)
+        tasks = sorted(tasks, key=lambda t: t.task_id)
+        batch_sz = len(tasks)
 
-        input_ids = torch.zeros(len(tasks), dtype=torch.long, device=self.device)
-        for i, task in enumerate(tasks):
-            if task.output_ids:
-                input_ids[i] = task.output_ids[-1]
-            else:
-                input_ids[i] = task.prompt_ids[-1]
+        for t in tasks:
+            self._maybe_alloc_page(t, start_pos)
 
-        input_tensor = input_ids.unsqueeze(1)
-        active_mask = torch.ones((len(tasks), 1), dtype=torch.bool, device=self.device)
+        input_ids = torch.tensor(
+            [t.output_ids[-1] if t.output_ids else t.prompt_ids[-1] for t in tasks],
+            dtype=torch.long,
+            device=self.device,
+        )
+
+        active_mask = torch.ones((batch_sz, 1), dtype=torch.bool, device=self.device)
+
+        page_tables = self._make_page_table_tensor(tasks)
+        total_len = start_pos + 1
+
+        temperatures = torch.tensor([t.temperature for t in tasks], device=self.device)
+        top_ks = torch.tensor([t.top_k for t in tasks], device=self.device)
+        top_ps = torch.tensor([t.top_p for t in tasks], device=self.device)
 
         with torch.inference_mode():
             outputs = self.model(
-                input_tensor,
+                input_ids.unsqueeze(1),
                 input_mask=active_mask,
-                persistent_key_values=self.kv_cache,
+                paged_cache=self.page_cache.bind(page_tables, total_len=total_len),
                 start_pos=start_pos,
             )
             logits = outputs["logits"][:, -1, :]
 
-        next_token_ids = []
-        for i, task in enumerate(tasks):
-            logit = logits[i : i + 1]
-            logit = apply_sampling_strategies(
-                logit,
-                task.temperature,
-                task.top_k,
-                task.top_p,
-            )
-            probs = torch.softmax(logit, dim=-1)
-            next_token = torch.multinomial(probs, num_samples=1)
-            next_token_ids.append(next_token.item())
+        next_tokens = sample(
+            logits,
+            temperature=temperatures,
+            top_k=top_ks,
+            top_p=top_ps,
+        ).tolist()
 
-        for task, next_token in zip(tasks, next_token_ids):
-            task.output_ids.append(next_token)
-            task.output_tokens += 1
+        for t, ntok in zip(tasks, next_tokens):
+            t.output_ids.append(ntok)
+            t.output_tokens += 1
+            pos = t.input_tokens + t.output_tokens
+            self._maybe_alloc_page(t, pos)
+            if t.stream_callback:
+                t.stream_callback(self.tokenizer.decode([ntok]))
 
-            pos = task.input_tokens + task.output_tokens
-            if task.slot >= 0 and pos < self.max_seq_len:
-                self.seq_mask[task.slot, pos] = True
+        for t in tasks:
+            if t.is_finished(self.tokenizer.stop_ids):
+                if t.stream_callback:
+                    t.stream_callback(STOP)
 
-            if task.stream_callback:
-                token_str = self.tokenizer.decode([next_token])
-                task.stream_callback(token_str)
+    def _make_page_table_tensor(self, tasks: List[Task]) -> Tensor:
+        max_pages = max(t.n_pages for t in tasks)
+        rows = [t.page_table + [-1] * (max_pages - t.n_pages) for t in tasks]
+        return torch.tensor(rows, dtype=torch.long, device=self.device)
 
-        for task in tasks:
-            if task.output_tokens >= task.max_tokens or (
-                task.output_ids and task.output_ids[-1] in self.tokenizer.stop_ids
-            ):
-                if task.stream_callback:
-                    task.stream_callback("[DONE]")
+    def _maybe_alloc_page(self, task: Task, pos: int) -> None:
+        needed = self._n_pages_for(pos + 1)
+        while task.n_pages < needed:
+            p = self.page_cache.alloc()
+            if p < 0:
+                break
+            task.page_table.append(p)
+            task.n_pages += 1
 
     def _run_generation_loop(self) -> None:
-        """Main generation loop."""
-        while self._running:
-            self._remove_finished_tasks()
-            self._refill_active_batch()
+        try:
+            while self._running:
+                self._remove_finished_tasks()
+                self._refill_active_batch()
 
-            if not self.active_tasks:
-                self._task_event.wait(timeout=0.01)
-                self._task_event.clear()
-                continue
+                if not self.active_tasks and not self.waiting_queue:
+                    self._task_event.clear()
+                    self._task_event.wait(timeout=1.0)
+                    continue
 
-            new_tasks = [t for t in self.active_tasks if t.output_tokens == 0]
-            decode_tasks = [t for t in self.active_tasks if t.output_tokens > 0]
+                to_prefill = [t for t in self.active_tasks if t.output_tokens == 0]
+                if to_prefill:
+                    for t in to_prefill:
+                        t.input_tokens = len(t.prompt_ids)
 
-            if decode_tasks:
-                start_pos = max(t.input_tokens + t.output_tokens for t in decode_tasks)
-            else:
-                start_pos = 0
+                    groups: Dict[Tuple[int, int], List[Task]] = {}
+                    for t in to_prefill:
+                        key = (len(t.prompt_ids), t._prefix_cached_tokens)
+                        groups.setdefault(key, []).append(t)
 
-            if new_tasks:
-                self._execute_prefill(new_tasks)
-                decode_tasks = new_tasks
-                start_pos = max(t.input_tokens for t in decode_tasks)
+                    for (prompt_len, start_pos), group in groups.items():
+                        if start_pos < prompt_len:
+                            self._execute_prefill(group, prompt_len, start_pos)
 
-            if decode_tasks:
-                self._execute_decode(decode_tasks, start_pos)
+                pos_groups: Dict[int, List[Task]] = {}
+                for t in self.active_tasks:
+                    pos_groups.setdefault(t.next_pos, []).append(t)
 
-            if not self.active_tasks and not self.waiting_queue:
-                self._task_event.wait(timeout=0.05)
-                self._task_event.clear()
+                if pos_groups:
+                    best_pos = max(pos_groups, key=lambda p: len(pos_groups[p]))
+                    self._execute_decode(pos_groups[best_pos], best_pos)
+        except Exception as e:
+            logger.error(f"Scheduler loop crashed: {e}", exc_info=True)
+            for task in self.active_tasks:
+                if task.stream_callback:
+                    task.stream_callback(STOP)
+            for task in self.waiting_queue:
+                if task.stream_callback:
+                    task.stream_callback(STOP)
+            raise
 
     def start(self) -> None:
-        """Start the generation loop."""
         if not self._running:
             self._running = True
-            self._loop_thread = threading.Thread(target=self._run_generation_loop)
-            self._loop_thread.daemon = True
-            self._loop_thread.start()
+            t = threading.Thread(target=self._run_generation_loop, daemon=True)
+            t.start()
+            self._loop_thread = t
 
     def stop(self) -> None:
-        """Stop the generation loop."""
         self._running = False
+        self._task_event.set()
         if hasattr(self, "_loop_thread"):
-            self._loop_thread.join(timeout=1.0)
-
-        # Clear KV cache to free GPU memory
-        if self.kv_cache is not None:
-            k_cache, v_cache = self.kv_cache
-            if k_cache is not None:
-                k_cache.detach()
-            if v_cache is not None:
-                v_cache.detach()
-
-        # Clear seq mask
-        self.seq_mask.detach()
-
-        # Clear task lists
+            self._loop_thread.join(timeout=2.0)
         self.waiting_queue.clear()
         self.active_tasks.clear()
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
 
     def get_stats(self) -> Dict[str, Any]:
-        """Get scheduler statistics."""
         return {
             "total_tasks": self._total_tasks,
             "total_tokens": self._total_tokens,

astrai/inference/server.py

@@ -1,15 +1,14 @@
 """
-Inference Server with Continuous Batching Support
-
-FastAPI server for inference with continuous batching.
-Provides OpenAI-compatible chat completion endpoints.
+OpenAI / Anthropic-compatible chat completion server backed by continuous-batching inference.
 """
 
 import json
 import logging
+import time
+import uuid
 from contextlib import asynccontextmanager
 from pathlib import Path
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Union
 
 import torch
 import uvicorn
@@ -23,18 +22,63 @@ from astrai.tokenize import AutoTokenizer
 
 logger = logging.getLogger(__name__)
 
-# Global model parameter and engine (loaded once)
-_engine: Optional[InferenceEngine] = None
-_model_param: Optional[Any] = None
 _project_root = Path(__file__).parent.parent.parent
 
-# Server configuration (set before running server)
-_server_config: Dict[str, Any] = {
-    "device": "cuda",
-    "dtype": torch.bfloat16,
-    "param_path": None,
-    "max_batch_size": 16,
-}
+
+class ServerState:
+    def __init__(self):
+        self.engine: Optional[InferenceEngine] = None
+        self.config: Dict[str, Any] = {
+            "device": "cuda",
+            "dtype": torch.bfloat16,
+            "param_path": None,
+            "max_batch_size": 16,
+        }
+
+
+_state = ServerState()
+
+
+class ChatMessage(BaseModel):
+    role: str
+    content: str
+
+
+class ChatCompletionRequest(BaseModel):
+    """OpenAI Chat Completion API request body."""
+
+    model: str = "astrai"
+    messages: List[ChatMessage]
+    temperature: Optional[float] = Field(default=1.0, ge=0.0, le=2.0)
+    top_p: Optional[float] = Field(default=1.0, ge=0.0, le=1.0)
+    top_k: Optional[int] = Field(default=50, ge=1)
+    stream: Optional[bool] = False
+    stop: Optional[Union[str, List[str]]] = None
+    max_tokens: Optional[int] = Field(default=2048, ge=1)
+    n: Optional[int] = Field(default=1, ge=1)
+    presence_penalty: Optional[float] = Field(default=0.0, ge=-2.0, le=2.0)
+    frequency_penalty: Optional[float] = Field(default=0.0, ge=-2.0, le=2.0)
+    logit_bias: Optional[Dict[int, float]] = None
+    user: Optional[str] = None
+
+
+class AnthropicMessage(BaseModel):
+    role: str
+    content: Union[str, List[Dict[str, Any]]]
+
+
+class MessagesRequest(BaseModel):
+    """Anthropic Messages API request body."""
+
+    model: str = "astrai"
+    max_tokens: int = Field(default=1024, ge=1)
+    messages: List[AnthropicMessage]
+    system: Optional[str] = None
+    temperature: Optional[float] = Field(default=1.0, ge=0.0, le=2.0)
+    top_p: Optional[float] = Field(default=1.0, ge=0.0, le=1.0)
+    top_k: Optional[int] = Field(default=50, ge=1)
+    stream: Optional[bool] = False
+    stop_sequences: Optional[List[str]] = None
 
 
 def configure_server(
@@ -43,39 +87,29 @@ def configure_server(
     param_path: Optional[Path] = None,
     max_batch_size: int = 16,
 ):
-    """Configure server settings before starting.
-
-    Args:
-        device: Device to load model on (e.g., "cuda", "cpu", "cuda:0")
-        dtype: Data type for model weights (e.g., torch.bfloat16, torch.float16)
-        param_path: Path to model parameters directory
-        max_batch_size: Maximum batch size for continuous batching
-    """
-    _server_config["device"] = device
-    _server_config["dtype"] = dtype
-    _server_config["param_path"] = param_path
-    _server_config["max_batch_size"] = max_batch_size
+    _state.config.update(
+        device=device,
+        dtype=dtype,
+        param_path=param_path,
+        max_batch_size=max_batch_size,
+    )
 
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    """Lifespan context manager for startup and shutdown events."""
-    global _model_param, _engine
-    # Startup: Load model with configured settings
     try:
         load_model(
-            param_path=_server_config["param_path"],
-            device=_server_config["device"],
-            dtype=_server_config["dtype"],
-            max_batch_size=_server_config["max_batch_size"],
+            param_path=_state.config["param_path"],
+            device=_state.config["device"],
+            dtype=_state.config["dtype"],
+            max_batch_size=_state.config["max_batch_size"],
         )
     except Exception as e:
         logger.error(f"Failed to load model: {e}")
         raise
     yield
-    # Shutdown: Cleanup engine
-    if _engine:
-        _engine.shutdown()
+    if _state.engine:
+        _state.engine.shutdown()
         logger.info("Inference engine shutdown complete")
 
 
@@ -88,203 +122,345 @@ def load_model(
     dtype: torch.dtype = torch.bfloat16,
     max_batch_size: int = 16,
 ):
-    """Load model parameters and initialize inference engine."""
-    global _model_param, _engine
     if param_path is None:
         param_path = _project_root / "params"
     if not param_path.exists():
         raise FileNotFoundError(f"Parameter directory not found: {param_path}")
 
-    # Load tokenizer separately
     tokenizer = AutoTokenizer.from_pretrained(param_path)
-    _model_param = AutoModel.from_pretrained(param_path)
-    _model_param.to(device=device, dtype=dtype)
+    model = AutoModel.from_pretrained(param_path)
+    model.to(device=device, dtype=dtype)
     logger.info(f"Model loaded on {device} with dtype {dtype}")
 
-    # Initialize inference engine with separate model and tokenizer
-    _engine = InferenceEngine(
-        model=_model_param,
+    _state.engine = InferenceEngine(
+        model=model,
         tokenizer=tokenizer,
         max_batch_size=max_batch_size,
     )
     logger.info(f"Inference engine initialized with max_batch_size={max_batch_size}")
 
 
-# Pydantic models for API request/response
-class ChatMessage(BaseModel):
-    role: str  # "user", "assistant", "system"
-    content: str
+def _get_engine() -> InferenceEngine:
+    if _state.engine is None:
+        raise HTTPException(status_code=503, detail="Engine not initialized")
+    return _state.engine
 
 
-class ChatCompletionRequest(BaseModel):
-    messages: List[ChatMessage]
-    temperature: float = Field(0.8, ge=0.0, le=2.0)
-    top_p: float = Field(0.95, ge=0.0, le=1.0)
-    top_k: int = Field(50, ge=0)
-    max_tokens: int = Field(2048, ge=1)
-    stream: bool = False
-    system_prompt: Optional[str] = None
-
-
-class CompletionResponse(BaseModel):
-    id: str = "chatcmpl-default"
-    object: str = "chat.completion"
-    created: int = 0
-    model: str = "astrai"
-    choices: List[Dict[str, Any]]
+def _make_chunk(
+    delta: Dict[str, str],
+    finish_reason: Optional[str] = None,
+    *,
+    resp_id: str,
+    created: int,
+    model: str,
+    index: int = 0,
+) -> str:
+    """Build a single SSE ``data:`` chunk matching OpenAI streaming format."""
+    data = {
+        "id": resp_id,
+        "object": "chat.completion.chunk",
+        "created": created,
+        "model": model,
+        "choices": [
+            {
+                "index": index,
+                "delta": delta,
+                "finish_reason": finish_reason,
+            }
+        ],
+    }
+    return f"data: {json.dumps(data, ensure_ascii=False)}\n\n"
 
 
 @app.get("/health")
 async def health():
     return {
         "status": "ok",
-        "model_loaded": _model_param is not None,
-        "engine_ready": _engine is not None,
+        "model_loaded": _state.engine is not None,
     }
 
 
 @app.get("/stats")
 async def get_stats():
-    """Get inference engine statistics."""
-    if _engine is None:
-        raise HTTPException(status_code=503, detail="Engine not initialized")
-    return _engine.get_stats()
+    return _get_engine().get_stats()
 
 
-@app.post("/v1/chat/completions", response_model=CompletionResponse)
+@app.post("/v1/chat/completions")
 async def chat_completion(request: ChatCompletionRequest):
-    """OpenAI-compatible chat completion endpoint.
+    """OpenAI-compatible chat completion endpoint (streaming + non-streaming)."""
+    engine = _get_engine()
+    resp_id = f"chatcmpl-{uuid.uuid4().hex[:12]}"
+    created = int(time.time())
+    model = request.model
 
-    Supports both streaming and non-streaming modes with continuous batching.
-    """
-    if _engine is None:
-        raise HTTPException(status_code=503, detail="Engine not initialized")
-
-    # Convert messages to prompt using engine's tokenizer
-    # Extract system prompt if present, then apply chat template
-    # Apply chat template directly with messages
-    prompt = _engine.tokenizer.apply_chat_template(
+    prompt = engine.tokenizer.apply_chat_template(
         [{"role": m.role, "content": m.content} for m in request.messages],
         tokenize=False,
     )
+    prompt_tokens = len(engine.tokenizer.encode(prompt))
 
     if request.stream:
-        # Streaming response (use synchronous generator)
-        generator = _engine.generate(
+        agen = engine.generate_async(
             prompt=prompt,
-            stream=True,
             max_tokens=request.max_tokens,
             temperature=request.temperature,
             top_p=request.top_p,
             top_k=request.top_k,
         )
 
-        def generate_stream():
-            for token in generator:
-                if token == "[DONE]":
-                    break
-                yield f"data: {json.dumps({'choices': [{'delta': {'content': token}}]})}\n\n"
+        async def event_stream():
+            yield _make_chunk(
+                {"role": "assistant"},
+                finish_reason=None,
+                resp_id=resp_id,
+                created=created,
+                model=model,
+            )
+
+            completion_tokens = 0
+            async for token in agen:
+                yield _make_chunk(
+                    {"content": token},
+                    finish_reason=None,
+                    resp_id=resp_id,
+                    created=created,
+                    model=model,
+                )
+                completion_tokens += 1
+
+            yield _make_chunk(
+                {},
+                finish_reason="stop",
+                resp_id=resp_id,
+                created=created,
+                model=model,
+            )
+
+            usage = {
+                "prompt_tokens": prompt_tokens,
+                "completion_tokens": completion_tokens,
+                "total_tokens": prompt_tokens + completion_tokens,
+            }
+            yield f"data: {json.dumps(usage, ensure_ascii=False)}\n\n"
             yield "data: [DONE]\n\n"
 
         return StreamingResponse(
-            generate_stream(),
+            event_stream(),
             media_type="text/event-stream",
             headers={"Cache-Control": "no-cache", "Connection": "keep-alive"},
         )
-    else:
-        # Non-streaming response
-        result = _engine.generate(
+
+    completion_tokens = 0
+    chunks: List[str] = []
+    agen = engine.generate_async(
+        prompt=prompt,
+        max_tokens=request.max_tokens,
+        temperature=request.temperature,
+        top_p=request.top_p,
+        top_k=request.top_k,
+    )
+    async for token in agen:
+        chunks.append(token)
+        completion_tokens += 1
+    content = "".join(chunks)
+
+    return {
+        "id": resp_id,
+        "object": "chat.completion",
+        "created": created,
+        "model": model,
+        "choices": [
+            {
+                "index": 0,
+                "message": {"role": "assistant", "content": content},
+                "finish_reason": "stop",
+            }
+        ],
+        "usage": {
+            "prompt_tokens": prompt_tokens,
+            "completion_tokens": completion_tokens,
+            "total_tokens": prompt_tokens + completion_tokens,
+        },
+    }
+
+
+def _make_anthropic_sse(event: str, data: Dict[str, Any]) -> str:
+    return f"event: {event}\ndata: {json.dumps(data, ensure_ascii=False)}\n\n"
+
+
+def _check_stop_sequence(text: str, stop_sequences: List[str]) -> Optional[str]:
+    for seq in stop_sequences:
+        if seq and seq in text:
+            return seq
+    return None
+
+
+def _extract_text_content(content: Union[str, List[Dict[str, Any]]]) -> str:
+    if isinstance(content, str):
+        return content
+    if isinstance(content, list):
+        for block in content:
+            if isinstance(block, dict) and block.get("type") == "text":
+                return block.get("text", "")
+    return ""
+
+
+def _build_anthropic_messages(
+    messages: List[AnthropicMessage], system: Optional[str]
+) -> List[Dict[str, str]]:
+    result: List[Dict[str, str]] = []
+    if system:
+        result.append({"role": "system", "content": system})
+    for m in messages:
+        content = _extract_text_content(m.content)
+        if content:
+            result.append({"role": m.role, "content": content})
+    return result
+
+
+@app.post("/v1/messages")
+async def create_message(request: MessagesRequest):
+    """Anthropic-compatible Messages API endpoint (streaming + non-streaming)."""
+    engine = _get_engine()
+    resp_id = f"msg_{uuid.uuid4().hex[:24]}"
+    model = request.model
+
+    chat_messages = _build_anthropic_messages(request.messages, request.system)
+    prompt = engine.tokenizer.apply_chat_template(chat_messages, tokenize=False)
+    prompt_tokens = len(engine.tokenizer.encode(prompt))
+
+    stop_sequences = request.stop_sequences or []
+
+    if request.stream:
+        agen = engine.generate_async(
             prompt=prompt,
-            stream=False,
             max_tokens=request.max_tokens,
             temperature=request.temperature,
             top_p=request.top_p,
             top_k=request.top_k,
         )
 
-        # Build OpenAI-style response
-        import time
-
-        resp = CompletionResponse(
-            id=f"chatcmpl-{int(time.time())}",
-            created=int(time.time()),
-            choices=[
+        async def event_stream():
+            yield _make_anthropic_sse(
+                "message_start",
                 {
+                    "type": "message_start",
+                    "message": {
+                        "id": resp_id,
+                        "type": "message",
+                        "role": "assistant",
+                        "model": model,
+                        "content": [],
+                        "usage": {"input_tokens": prompt_tokens},
+                    },
+                },
+            )
+
+            yield _make_anthropic_sse(
+                "content_block_start",
+                {
+                    "type": "content_block_start",
                     "index": 0,
-                    "message": {"role": "assistant", "content": result},
-                    "finish_reason": "stop",
-                }
-            ],
-        )
-        return resp
+                    "content_block": {"type": "text", "text": ""},
+                },
+            )
 
+            completion_tokens = 0
+            accumulated = ""
+            stopped_seq: Optional[str] = None
+            async for token in agen:
+                accumulated += token
+                completion_tokens += 1
 
-@app.post("/generate")
-async def generate(
-    query: str,
-    history: Optional[List[List[str]]] = None,
-    temperature: float = 0.8,
-    top_p: float = 0.95,
-    top_k: int = 50,
-    max_len: int = 2048,
-    stream: bool = False,
-):
-    """Simple generation endpoint.
+                matched = _check_stop_sequence(accumulated, stop_sequences)
+                if matched:
+                    text = accumulated[: accumulated.rfind(matched)]
+                    stopped_seq = matched
+                    if text:
+                        yield _make_anthropic_sse(
+                            "content_block_delta",
+                            {
+                                "type": "content_block_delta",
+                                "index": 0,
+                                "delta": {"type": "text_delta", "text": text},
+                            },
+                        )
+                    break
 
-    Args:
-        query: Input query string
-        history: Conversation history as list of [user, assistant] pairs
-        temperature: Sampling temperature
-        top_p: Top-p sampling parameter
-        top_k: Top-k sampling parameter
-        max_len: Maximum tokens to generate
-        stream: Enable streaming output
+                yield _make_anthropic_sse(
+                    "content_block_delta",
+                    {
+                        "type": "content_block_delta",
+                        "index": 0,
+                        "delta": {"type": "text_delta", "text": token},
+                    },
+                )
 
-    Returns:
-        dict: Generation result with response field
-    """
-    if _engine is None:
-        raise HTTPException(status_code=503, detail="Engine not initialized")
+            yield _make_anthropic_sse(
+                "content_block_stop",
+                {"type": "content_block_stop", "index": 0},
+            )
 
-    # Build messages for chat template
-    messages = []
-    if history:
-        # Convert history format: List[List[str]] -> List[Dict]
-        for h in history:
-            if len(h) >= 2:
-                messages.append({"role": "user", "content": h[0]})
-                messages.append({"role": "assistant", "content": h[1]})
-    messages.append({"role": "user", "content": query})
+            stop_reason = "stop_sequence" if stopped_seq else "end_turn"
+            yield _make_anthropic_sse(
+                "message_delta",
+                {
+                    "type": "message_delta",
+                    "delta": {"stop_reason": stop_reason, "stop_sequence": stopped_seq},
+                    "usage": {"output_tokens": completion_tokens},
+                },
+            )
 
-    # Use tokenizer's chat template
-    prompt = _engine.tokenizer.apply_chat_template(messages, tokenize=False)
+            yield _make_anthropic_sse(
+                "message_stop",
+                {"type": "message_stop"},
+            )
 
-    if stream:
-        # Synchronous streaming
-        result = _engine.generate(
-            prompt=prompt,
-            stream=True,
-            max_tokens=max_len,
-            temperature=temperature,
-            top_p=top_p,
-            top_k=top_k,
+        return StreamingResponse(
+            event_stream(),
+            media_type="text/event-stream",
+            headers={"Cache-Control": "no-cache", "Connection": "keep-alive"},
         )
 
-        def stream_generator():
-            for token in result:
-                yield token + "\n"
+    completion_tokens = 0
+    chunks: List[str] = []
+    agen = engine.generate_async(
+        prompt=prompt,
+        max_tokens=request.max_tokens,
+        temperature=request.temperature,
+        top_p=request.top_p,
+        top_k=request.top_k,
+    )
+    stopped_seq: Optional[str] = None
+    accumulated = ""
+    async for token in agen:
+        chunks.append(token)
+        completion_tokens += 1
+        accumulated += token
+        matched = _check_stop_sequence(accumulated, stop_sequences)
+        if matched:
+            stopped_seq = matched
+            break
 
-        return StreamingResponse(stream_generator(), media_type="text/plain")
-    else:
-        result = _engine.generate(
-            prompt=prompt,
-            stream=False,
-            max_tokens=max_len,
-            temperature=temperature,
-            top_p=top_p,
-            top_k=top_k,
-        )
-        return {"response": result}
+    content = "".join(chunks)
+    if stopped_seq:
+        idx = content.rfind(stopped_seq)
+        if idx != -1:
+            content = content[:idx]
+
+    return {
+        "id": resp_id,
+        "type": "message",
+        "role": "assistant",
+        "model": model,
+        "content": [{"type": "text", "text": content}],
+        "stop_reason": "stop_sequence" if stopped_seq else "end_turn",
+        "stop_sequence": stopped_seq,
+        "usage": {
+            "input_tokens": prompt_tokens,
+            "output_tokens": completion_tokens,
+        },
+    }
 
 
 def run_server(
@@ -296,17 +472,6 @@ def run_server(
     param_path: Optional[Path] = None,
     max_batch_size: int = 16,
 ):
-    """Run the FastAPI server with uvicorn.
-
-    Args:
-        host: Server host address
-        port: Server port number
-        reload: Enable auto-reload for development
-        device: Device to load model on (e.g., "cuda", "cpu", "cuda:0")
-        dtype: Data type for model weights (e.g., torch.bfloat16, torch.float16)
-        param_path: Path to model parameters directory
-        max_batch_size: Maximum batch size for continuous batching
-    """
     configure_server(
         device=device,
         dtype=dtype,

astrai/model/automodel.py

@@ -4,12 +4,13 @@ AutoModel base class for model loading and saving.
 
 from contextlib import contextmanager
 from pathlib import Path
-from typing import Dict, Self, Type, Union
+from typing import Self, Type, Union
 
 import safetensors.torch as st
 import torch.nn as nn
 
 from astrai.config import ModelConfig
+from astrai.factory import Registry
 
 
 @contextmanager
@@ -44,8 +45,7 @@ class AutoModel(nn.Module):
     Provides model loading/saving and generation capabilities.
     """
 
-    # Model registry - stored as class attribute
-    _registry: Dict[str, Type["AutoModel"]] = {}
+    _registry = Registry()
 
     def __init__(self, config: ModelConfig):
         super().__init__()
@@ -63,7 +63,7 @@ class AutoModel(nn.Module):
         """
 
         def decorator(sub_cls: Type["AutoModel"]) -> Type["AutoModel"]:
-            cls._registry[model_type.lower()] = sub_cls
+            cls._registry.register(model_type.lower(), sub_cls)
             return sub_cls
 
         return decorator
@@ -72,18 +72,19 @@ class AutoModel(nn.Module):
     def get_model_class(cls, model_type: str) -> Type["AutoModel"]:
         """Get model class by model_type string."""
         model_type = model_type.lower()
-        if model_type not in cls._registry:
-            available = list(cls._registry.keys())
+        if not cls._registry.contains(model_type):
+            available = cls._registry.list_names()
             raise ValueError(
                 f"Unknown model_type: {model_type}. Available: {available}"
             )
-        return cls._registry[model_type]
+        return cls._registry.get(model_type)
 
     @classmethod
     def from_pretrained(
         cls,
         path: Union[str, Path],
         disable_random_init: bool = True,
+        strict: bool = True,
     ) -> nn.Module:
 
         model_path = Path(path)
@@ -96,14 +97,8 @@ class AutoModel(nn.Module):
         else:
             raise FileNotFoundError(f"Config file not found: {config_path}")
 
-        # If called from base class, use model_type to determine actual model class
-        if cls is AutoModel:
-            model_type = config.model_type or "transformer"
-            actual_cls = cls.get_model_class(model_type)
-        else:
-            raise ValueError(
-                f"Cannot call from_pretrained() on subclass {cls.__name__}"
-            )
+        model_type = config.model_type or "transformer"
+        actual_cls = cls.get_model_class(model_type)
 
         with _disable_random_init(enable=disable_random_init):
             model = actual_cls(config)
@@ -112,7 +107,7 @@ class AutoModel(nn.Module):
         weights_path = model_path / "model.safetensors"
         if weights_path.exists():
             state_dict = st.load_file(str(weights_path))
-            model.load_state_dict(state_dict, strict=False)
+            model.load_state_dict(state_dict, strict=strict)
 
         return model
 

astrai/model/module.py

@@ -5,17 +5,11 @@ import torch.nn as nn
 import torch.nn.functional as F
 from torch import Tensor
 
+from astrai.inference.cache import CacheView
+
 
 def repeat_kv(x: Tensor, n_rep: int) -> Tensor:
-    """
-    Repeat k times along the dimension for attention heads.
-    Args:
-        x (Tensor): The input tensor.
-        n_rep (int): The number of repetitions.
-    Returns:
-        Tensor: The repeated tensor.
-    """
-
+    """Repeat KV heads n_rep times for GQA."""
     bs, slen, n_heads, head_dim = x.shape
     if n_rep == 1:
         return x
@@ -32,49 +26,25 @@ def get_rotary_emb(
     base: float = 10000,
     device: Optional[torch.device] = None,
 ) -> Tuple[Tensor, Tensor]:
-    """
-    Get the rotary embedding for the given dimension and maximum length.
-    Args:
-        dim (int): The dimension of the input.
-        max_len (int): The maximum length of the input.
-        base (float, optional): The base for the frequency. Defaults to 10000.
-        device (optional): The device to create tensors on. Defaults to None.
-    Returns:
-        Tensor: The rotary embedding tensor.
-    """
-
+    """Precompute cos/sin for RoPE."""
     theta = base ** (-torch.arange(0, dim, 2, dtype=torch.float64, device=device) / dim)
     t = torch.arange(0, max_len, dtype=torch.float64, device=device)
     freqs = torch.outer(t, theta)
-
     return torch.cos(freqs).float(), torch.sin(freqs).float()
 
 
 def apply_rotary_emb(x: torch.Tensor, rotary_emb: Tuple[Tensor, Tensor]) -> Tensor:
-    """
-    Apply rotary embedding to the input tensor using cos/sin form.
-    Args:
-        x (Tensor): The input tensor (shape [..., seq_len, dim]).
-        rotary_emb (Tuple[Tensor, Tensor]): The rotary embedding (shape [seq_len, dim//2]).
-    Returns:
-        Tensor: The output tensor (rotated, same shape as input).
-    """
-
+    """Apply rotary embedding via cos/sin (shape-preserving)."""
     dtype = x.dtype
     cos, sin = rotary_emb
-
-    cos = cos.unsqueeze(0).unsqueeze(2)  # [1, seq_len, 1, dim//2]
-    sin = sin.unsqueeze(0).unsqueeze(2)  # [1, seq_len, 1, dim//2]
-
-    x_real = x[..., 0::2]  # [batch, seq_len, dim//2]
-    x_imag = x[..., 1::2]  # [batch, seq_len, dim//2]
-
+    cos = cos.unsqueeze(0).unsqueeze(2)
+    sin = sin.unsqueeze(0).unsqueeze(2)
+    x_real = x[..., 0::2]
+    x_imag = x[..., 1::2]
     x_real_rot = x_real * cos - x_imag * sin
     x_imag_rot = x_real * sin + x_imag * cos
-
-    x_out = torch.stack([x_real_rot, x_imag_rot], dim=-1)  # [batch, seq_len, dim//2, 2]
-    x_out = x_out.view(*x_out.shape[:-2], -1)  # [batch, seq_len, dim]
-
+    x_out = torch.stack([x_real_rot, x_imag_rot], dim=-1)
+    x_out = x_out.view(*x_out.shape[:-2], -1)
     return x_out.to(dtype)
 
 
@@ -95,13 +65,10 @@ class RotaryEmbedding(nn.Module):
 
     def forward(self, x: Tensor, start_pos: int = 0) -> Tuple[Tensor, Tensor]:
         seq_len = x.size(1)
-
         if self.max_len_cached < seq_len + start_pos:
             self._set_rotary_buffer(self.max_len_cached * 2, x.device)
-
         cos = self.cos_cached[start_pos : start_pos + seq_len]
         sin = self.sin_cached[start_pos : start_pos + seq_len]
-
         return (cos, sin)
 
 
@@ -185,13 +152,13 @@ class GQA(nn.Module):
         x: Tensor,
         rotary_emb: Tuple[Tensor, Tensor],
         mask: Tensor = None,
-        kv_cache: Optional[Tuple[Tensor, Tensor]] = None,
+        paged_cache: Optional[CacheView] = None,
         start_pos: int = 0,
     ) -> Tensor:
         bsz, seq_len, _ = x.size()
         is_causal = mask is None
 
-        # x(bsz, seq_len, n_heads * head_dim) -> (bsz, seq_len, n_heads, head_dim)
+        # (bsz, seq_len, dim) -> (bsz, seq_len, n_heads, head_dim)
         q = self._split_heads(self.q_proj(x), self.n_heads)
         k = self._split_heads(self.k_proj(x), self.n_kv_heads)
         v = self._split_heads(self.v_proj(x), self.n_kv_heads)
@@ -200,22 +167,14 @@ class GQA(nn.Module):
         if self.use_qk_norm:
             q, k = self.q_norm(q), self.k_norm(k)
 
-        if kv_cache is not None:
-            k_cache, v_cache = kv_cache
-
-            # copy to cache
-            k_cache[:bsz, start_pos : start_pos + seq_len, self.layer_id] = k
-            v_cache[:bsz, start_pos : start_pos + seq_len, self.layer_id] = v
-
-            # get cache
-            k = k_cache[:bsz, : start_pos + seq_len, self.layer_id]
-            v = v_cache[:bsz, : start_pos + seq_len, self.layer_id]
+        if paged_cache is not None:
+            paged_cache.write(self.layer_id, start_pos, k, v)
+            k, v = paged_cache.gather(self.layer_id)
 
         k, v = repeat_kv(k, self.n_rep), repeat_kv(v, self.n_rep)
 
         # (bsz, seq_len, n_heads, head_dim) -> (bsz, n_heads, seq_len, head_dim)
         q, k, v = q.permute(0, 2, 1, 3), k.permute(0, 2, 1, 3), v.permute(0, 2, 1, 3)
-        # (bsz, n_heads, seq_len, head_dim) - > (bsz, seq_len, n_heads*head_dim)
         sdqa_out = (
             F.scaled_dot_product_attention(q, k, v, mask, is_causal=is_causal)
             .permute(0, 2, 1, 3)
@@ -227,7 +186,6 @@ class GQA(nn.Module):
             sdqa_out = sdqa_out * F.sigmoid(self.gate(x))
 
         out = self.o_proj(sdqa_out)
-
         return out
 
 
@@ -260,7 +218,7 @@ class MLA(nn.Module):
         self.kv_a_proj = Linear(dim, kv_lora_rank, bias=False)
         self.kv_norm = RMSNorm(kv_lora_rank, norm_eps)
 
-        # KV (k_nope, k_rope, v)
+        # fused KV: (k_nope, k_rope, v)
         self.kv_b_proj = Linear(
             kv_lora_rank,
             n_kv_heads * (self.head_dim + qk_rope_head_dim + self.head_dim),
@@ -276,7 +234,7 @@ class MLA(nn.Module):
         x: Tensor,
         rotary_emb: Tuple[Tensor, Tensor],
         mask: Tensor = None,
-        kv_cache: Optional[Tuple[Tensor, Tensor]] = None,
+        paged_cache: Optional[CacheView] = None,
         start_pos: int = 0,
     ) -> Tensor:
         bsz, seq_len, _ = x.size()
@@ -305,12 +263,9 @@ class MLA(nn.Module):
         q = torch.cat([q_nope, q_rope], dim=-1)
         k = torch.cat([k_nope, k_rope], dim=-1)
 
-        if kv_cache is not None:
-            k_cache, v_cache = kv_cache
-            k_cache[:bsz, start_pos : start_pos + seq_len, self.layer_id] = k
-            v_cache[:bsz, start_pos : start_pos + seq_len, self.layer_id] = v
-            k = k_cache[:bsz, : start_pos + seq_len, self.layer_id]
-            v = v_cache[:bsz, : start_pos + seq_len, self.layer_id]
+        if paged_cache is not None:
+            paged_cache.write(self.layer_id, start_pos, k, v)
+            k, v = paged_cache.gather(self.layer_id)
 
         q = q.permute(0, 2, 1, 3)
         k = k.permute(0, 2, 1, 3)
@@ -323,7 +278,6 @@ class MLA(nn.Module):
             attn_out = attn_out * F.sigmoid(self.gate(x))
 
         out = self.o_proj(attn_out)
-
         return out
 
 
@@ -358,18 +312,19 @@ class DecoderBlock(nn.Module):
         x: Tensor,
         rotary_emb: Tuple[Tensor, Tensor],
         attention_mask: Optional[Tensor] = None,
-        kv_cache: Optional[Tuple[Tensor, Tensor]] = None,
+        paged_cache: Optional[CacheView] = None,
         start_pos: int = 0,
     ) -> Tensor:
-        # attention
         attn_output = self.attention(
-            self.input_norm(x), rotary_emb, attention_mask, kv_cache, start_pos
+            self.input_norm(x),
+            rotary_emb,
+            attention_mask,
+            paged_cache,
+            start_pos,
         )
         x = attn_output + x
 
-        # feed forward
         x = self.mlp(self.post_attention_norm(x)) + x
-
         return x
 
 

astrai/model/transformer.py

@@ -1,10 +1,11 @@
-from typing import Any, Mapping, Optional, Tuple
+from typing import Any, Mapping, Optional
 
 import torch
 import torch.nn as nn
 from torch import Tensor
 
 from astrai.config.model_config import ModelConfig
+from astrai.inference.cache import CacheView
 from astrai.model.automodel import AutoModel
 from astrai.model.module import (
     DecoderBlock,
@@ -21,39 +22,25 @@ def process_attention_mask(
     start_pos: int = 0,
     is_causal: bool = False,
 ) -> Tensor:
-    """
-    Create attention mask for GQA
-    Args:
-        seq_mask (Tensor): A tensor indicating whether each position is valid or not.
-        input_tensor (Tensor): The input tensor.
-        start_pos (int): The starting position of the sequence.
-        is_causal (bool): Whether the attention is causal or not.
-    Returns:
-        Tensor: The attention mask tensor.
-    """
+    """Build 4D attention mask from 2D seq_mask, with optional causal masking."""
     device = input_tensor.device
     dtype = input_tensor.dtype
     seq_len = input_tensor.size(1)
 
     if seq_mask is None:
         if start_pos != 0:
-            # for single prompt chat
             seq_mask = torch.ones((1, seq_len), dtype=torch.bool, device=device)
         else:
             return None
 
     if seq_mask.dim() > 2:
-        # shape (bsz, seq_len) or (bsz,n_heads, seq_len, seq_len + start_pos)
-        # if ndim > 2, it's 4D tensor
         return seq_mask
 
     batch_size = seq_mask.size(0)
     seq_mask = seq_mask[:, : start_pos + seq_len].to(device=device, dtype=torch.bool)
-    # (bsz, start_pos + seq_len)
     expanded_mask = seq_mask.unsqueeze(1).expand(
         batch_size, seq_len, start_pos + seq_len
     )
-    # (bsz, seq_len, start_pos + seq_len)
 
     if is_causal:
         expanded_mask = torch.tril(expanded_mask, diagonal=start_pos)
@@ -62,16 +49,13 @@ def process_attention_mask(
     attention_mask = attention_mask.masked_fill_(
         ~expanded_mask, -torch.finfo(dtype).max / 2
     ).unsqueeze(1)
-    # (bsz, 1, seq_len, seq_len + start_pos)
 
     return attention_mask
 
 
 @AutoModel.register("transformer")
 class Transformer(AutoModel):
-    """
-    Transformer language model.
-    """
+    """Transformer language model with paged KV cache."""
 
     def __init__(self, config: ModelConfig):
         super().__init__(config)
@@ -114,18 +98,15 @@ class Transformer(AutoModel):
         lm_head_key = "lm_head.weight"
         embed_key = "embed_tokens.weight"
 
-        # Make a copy to avoid modifying the original state_dict
         state_dict = dict(state_dict)
 
         if self.config.tie_weight:
-            # same tensor
+            # same tensor for embed and lm_head
             if embed_key in state_dict:
                 state_dict[lm_head_key] = state_dict[embed_key]
         else:
-            # If lm_head.weight exists in checkpoint, use it directly
-            # If not, copy from embed_tokens.weight
             if lm_head_key not in state_dict and embed_key in state_dict:
-                # use clone to avoid sharing the same tensor
+                # clone to avoid sharing gradients
                 state_dict[lm_head_key] = torch.clone(state_dict[embed_key])
 
         return super().load_state_dict(state_dict, strict, assign)
@@ -146,7 +127,7 @@ class Transformer(AutoModel):
         self,
         input_ids: Tensor,
         input_mask: Optional[Tensor] = None,
-        persistent_key_values: Optional[Tuple[Tensor, Tensor]] = None,
+        paged_cache: Optional[CacheView] = None,
         start_pos: int = 0,
     ) -> Tensor:
         assert input_ids.ndim == 2
@@ -157,7 +138,7 @@ class Transformer(AutoModel):
         attn_mask = process_attention_mask(input_mask, x, start_pos, is_causal=True)
 
         for layer in self.layers:
-            x = layer(x, rotary_emb, attn_mask, persistent_key_values, start_pos)
+            x = layer(x, rotary_emb, attn_mask, paged_cache, start_pos)
 
         hidden_states = self.norm(x)
         logits = self.lm_head(hidden_states)

astrai/parallel/setup.py

@@ -1,7 +1,7 @@
 import os
 from contextlib import contextmanager
 from functools import wraps
-from typing import Callable, List, Optional
+from typing import Callable
 
 import torch
 import torch.distributed as dist
@@ -34,7 +34,6 @@ def setup_parallel(
     master_addr: str = "localhost",
     master_port: str = "29500",
     device_type: str = "cuda",
-    device_ids: Optional[List[int]] = None,
 ):
 
     if dist.is_available() and dist.is_initialized():
@@ -45,15 +44,10 @@ def setup_parallel(
         yield None
         return
 
-    if device_ids is None:
-        device_ids = [i for i in range(world_size)]
-
-    rank = device_ids[rank % len(device_ids)]
-    device_id = torch.device(device_type, device_ids[rank])
+    device_id = torch.device(device_type, rank)
 
     os.environ["MASTER_ADDR"] = master_addr
     os.environ["MASTER_PORT"] = master_port
-
     os.environ["LOCAL_RANK"] = str(rank)
     os.environ["WORLD_SIZE"] = str(world_size)
     os.environ["LOCAL_DEVICE"] = str(device_id)
@@ -103,7 +97,6 @@ def wrapper_spawn_func(
     master_addr: str,
     master_port: str,
     device_type: str,
-    device_ids: List[int],
     func: Callable,
     kwargs: dict,
 ):
@@ -115,7 +108,6 @@ def wrapper_spawn_func(
             master_addr=master_addr,
             master_port=master_port,
             device_type=device_type,
-            device_ids=device_ids,
         ):
             func(**kwargs)
 
@@ -131,7 +123,6 @@ def spawn_parallel_fn(
     master_addr: str = "localhost",
     master_port: str = "29500",
     device_type: str = "cuda",
-    device_ids: Optional[List[int]] = None,
     **kwargs,
 ):
     # clear environment variables
@@ -147,8 +138,9 @@ def spawn_parallel_fn(
             del os.environ[key]
 
     if world_size == 1:
-        device_ids = device_ids or [0]
-        device_id = torch.device(device_type, device_ids[0])
+        device_id = torch.device(device_type, 0)
+        os.environ["LOCAL_RANK"] = "0"
+        os.environ["WORLD_SIZE"] = "1"
         os.environ["LOCAL_DEVICE"] = str(device_id)
 
         func(**kwargs)
@@ -160,7 +152,6 @@ def spawn_parallel_fn(
         master_addr,
         master_port,
         device_type,
-        device_ids,
         func,
         kwargs,
     )

astrai/serialization.py

@@ -1,7 +1,7 @@
 import json
 import os
 from pathlib import Path
-from typing import Any, Dict, List
+from typing import Any, Dict, List, Optional
 
 import h5py
 import safetensors.torch as st
@@ -54,10 +54,12 @@ class Checkpoint:
         state_dict: Dict[str, Any],
         epoch: int = 0,
         iteration: int = 0,
+        extra: Optional[Dict[str, Any]] = None,
     ):
         self.state_dict = state_dict
         self.epoch = epoch
         self.iteration = iteration
+        self.extra = extra or {}
 
     def save(
         self,
@@ -77,6 +79,8 @@ class Checkpoint:
                 json.dump(meta, f, indent=2)
 
             st.save_file(self.state_dict, save_path / "state_dict.safetensors")
+            if self.extra:
+                torch.save(self.extra, save_path / "extra.pt")
 
     @classmethod
     def load(
@@ -99,8 +103,14 @@ class Checkpoint:
 
         state_dict = st.load_file(save_path / "state_dict.safetensors")
 
+        extra = None
+        extra_path = save_path / "extra.pt"
+        if extra_path.exists():
+            extra = torch.load(extra_path, map_location="cpu", weights_only=False)
+
         return cls(
             state_dict=state_dict,
             epoch=meta["epoch"],
             iteration=meta["iteration"],
+            extra=extra,
         )

astrai/tokenize/tokenizer.py

@@ -64,6 +64,11 @@ class AutoTokenizer:
             save_path: Path to save the tokenizer
         """
 
+        if self._tokenizer is None:
+            raise RuntimeError(
+                "Tokenizer not initialized. Load or create a tokenizer first."
+            )
+
         save_path = Path(save_path)
         save_path.mkdir(parents=True, exist_ok=True)
 

astrai/trainer/strategy.py

@@ -265,7 +265,9 @@ class DPOStrategy(BaseStrategy):
 class GRPOStrategy(BaseStrategy):
     """Group Relative Policy Optimization strategy.
 
-    Implements GRPO with clipping and KL penalty.
+    On-policy GRPO following DeepSeek-R1: the policy model is updated while
+    a frozen ref_model stores the old-policy log-probs.  ratio = exp(logπ_θ - logπ_ref),
+    clipped PPO objective.  Call ``sync_ref_model()`` after each data-generation round.
     """
 
     def __init__(
@@ -276,6 +278,7 @@ class GRPOStrategy(BaseStrategy):
         kl_coef: float = 0.01,
         group_size: int = 4,
         reduction: str = "mean",
+        sync_interval: int = 200,
         **kwargs,
     ):
         super().__init__(model, device, **kwargs)
@@ -284,8 +287,19 @@ class GRPOStrategy(BaseStrategy):
         self.kl_coef = kl_coef
         self.group_size = group_size
         self.reduction = reduction
+        self.sync_interval = sync_interval
+        self._step = 0
+
+    def sync_ref_model(self):
+        """Copy current model weights to ref model."""
+        ref_state = self.model.state_dict()
+        self.ref_model.load_state_dict(ref_state)
 
     def compute_loss(self, batch: Dict[str, Tensor]) -> Tensor:
+        self._step += 1
+        if self._step % self.sync_interval == 0:
+            self.sync_ref_model()
+
         batch = move_to_device(batch, self.device)
         prompts = batch["prompts"]
         responses = batch["responses"]
@@ -297,7 +311,6 @@ class GRPOStrategy(BaseStrategy):
         masks_flat = masks.view(-1, response_len)
         prompt_expanded = prompts.unsqueeze(1).repeat(1, group_size, 1).flatten(0, 1)
 
-        # Shape: (batch_size * group_size, seq_len + response_len)
         full_sequences = torch.cat([prompt_expanded, responses_flat], dim=-1)
         full_masks = torch.cat([torch.ones_like(prompt_expanded), masks_flat], dim=-1)
 
@@ -312,14 +325,13 @@ class GRPOStrategy(BaseStrategy):
             )
             log_probs_ref = log_probs_ref.view(batch_size, group_size)
 
-        # Compute advantages from rewards with normalization
         eps = torch.finfo(log_probs_policy.dtype).eps
         mean = rewards.mean(dim=-1, keepdim=True)
         std = rewards.std(dim=-1, keepdim=True)
         advantages = (rewards - mean) / (std + eps)
 
-        # PPO-style clipped surrogate objective
-        ratio = torch.exp(0)  # Off-policy: policy_model = old_model
+        ratio = torch.exp(log_probs_policy - log_probs_ref)
+
         surr1 = ratio * advantages
         surr2 = torch.clamp(ratio, 1 - self.clip_eps, 1 + self.clip_eps) * advantages
 

astrai/trainer/train_callback.py

@@ -121,11 +121,13 @@ class CheckpointCallback(TrainCallback):
         interval: int,
         weight_only: bool = False,
         state_dict_fn: Optional[Callable[[nn.Module], dict]] = None,
+        save_extra_fn: Optional[Callable[["TrainContext"], dict]] = None,
     ):
         self.save_dir = save_dir
         self.interval = interval
         self.weight_only = weight_only
         self.state_dict_fn = state_dict_fn
+        self.save_extra_fn = save_extra_fn
         self.last_ckpt_iter = 0
 
     @only_on_rank(0)
@@ -139,8 +141,12 @@ class CheckpointCallback(TrainCallback):
             else context.model.state_dict()
         )
 
+        extra = self.save_extra_fn(context) if self.save_extra_fn else None
         context.checkpoint = Checkpoint(
-            state_dict=state_dict, epoch=context.epoch, iteration=context.iteration
+            state_dict=state_dict,
+            epoch=context.epoch,
+            iteration=context.iteration,
+            extra=extra,
         )
 
         context.checkpoint.save(save_path)

astrai/trainer/train_context.py

@@ -1,5 +1,5 @@
 from dataclasses import dataclass, field
-from typing import Optional, Self
+from typing import Callable, Optional, Self
 
 import torch.nn as nn
 from torch.optim import Optimizer
@@ -32,68 +32,70 @@ class TrainContext:
 
 
 class TrainContextBuilder:
-    def __init__(self, config: TrainConfig):
+    def __init__(
+        self,
+        config: TrainConfig,
+        load_extra_fn: Optional[Callable[[dict, "TrainContext"], None]] = None,
+    ):
         self.config = config
-        self._context = TrainContext(
-            model=config.model,
+        self._checkpoint: Optional[Checkpoint] = None
+        self._load_extra_fn = load_extra_fn
+
+    def with_checkpoint(self, checkpoint: Optional[Checkpoint]) -> Self:
+        self._checkpoint = checkpoint
+        return self
+
+    def build(self) -> TrainContext:
+        context = TrainContext(
+            model=self.config.model,
             world_size=get_world_size(),
             rank=get_rank(),
         )
 
         device = get_current_device()
-        self._context.model = self._context.model.to(device=device)
+        context.model = context.model.to(device=device)
 
-        if self.config.nprocs > 1:
-            fn = self.config.parallel_wrapper
-            self._context.model = fn(self._context.model)
+        if self.config.nprocs > 1 and self.config.parallel_wrapper:
+            context.model = self.config.parallel_wrapper(context.model)
 
-        self._context.optimizer = self.config.optimizer_fn(self._context.model)
-        self._context.scheduler = self.config.scheduler_fn(self._context.optimizer)
-
-    def with_checkpoint(self, checkpoint: Optional[Checkpoint]) -> Self:
-        if checkpoint is None:
-            checkpoint = Checkpoint(
-                state_dict=self._context.model.state_dict(),
-            )
+        if self._checkpoint is not None:
+            context.epoch = max(self._checkpoint.epoch, self.config.start_epoch)
+            context.iteration = max(self._checkpoint.iteration, self.config.start_batch)
+            context.model.load_state_dict(self._checkpoint.state_dict)
+            context.checkpoint = self._checkpoint
         else:
-            # resume from the assigned checkpoint or assigned iteration
-            self._context.epoch = max(checkpoint.epoch, self.config.start_epoch)
-            self._context.iteration = max(checkpoint.iteration, self.config.start_batch)
-            self._context.model.load_state_dict(checkpoint.state_dict)
+            context.checkpoint = Checkpoint(
+                state_dict=context.model.state_dict(),
+            )
 
-        self._context.checkpoint = checkpoint
-        return self
+        context.optimizer = self.config.optimizer_fn(context.model)
+        context.scheduler = self.config.scheduler_fn(context.optimizer)
 
-    def with_dataloader(self) -> Self:
-        # fix: change batch level iteration to sample level offset
-        config = self.config
-        sampler_offset = self._context.iteration * config.batch_size
-        resumeable_sampler = ResumableDistributedSampler(
-            data_source=config.dataset,
-            start_epoch=self._context.epoch,
+        if self._checkpoint and self._checkpoint.extra and self._load_extra_fn:
+            self._load_extra_fn(self._checkpoint.extra, context)
+
+        cfg = self.config
+        sampler_offset = context.iteration * cfg.batch_size
+        sampler = ResumableDistributedSampler(
+            data_source=cfg.dataset,
+            start_epoch=context.epoch,
             start_iter=sampler_offset,
-            seed=config.random_seed,
+            seed=cfg.random_seed,
+        )
+        context.dataloader = DataLoader(
+            cfg.dataset,
+            batch_size=cfg.batch_size,
+            sampler=sampler,
+            num_workers=cfg.num_workers,
+            pin_memory=cfg.pin_memory,
+            prefetch_factor=cfg.prefetch_factor,
         )
 
-        dataloader = DataLoader(
-            config.dataset,
-            batch_size=config.batch_size,
-            sampler=resumeable_sampler,
-            num_workers=config.num_workers,
-            pin_memory=config.pin_memory,
-            prefetch_factor=config.prefetch_factor,
-        )
-        self._context.dataloader = dataloader
-        return self
-
-    def with_strategy(self) -> Self:
-        self._context.strategy = StrategyFactory.create(
-            model=self._context.model,
+        context.strategy = StrategyFactory.create(
+            model=context.model,
             train_type=self.config.strategy,
-            device=get_current_device(),
+            device=device,
             **self.config.extra_kwargs,
         )
-        return self
 
-    def build(self) -> TrainContext:
-        return self._context
+        return context

astrai/trainer/trainer.py

@@ -35,11 +35,7 @@ class Trainer:
 
     def _build_context(self, checkpoint: Optional[Checkpoint]) -> TrainContext:
         return (
-            TrainContextBuilder(self.train_config)
-            .with_checkpoint(checkpoint)
-            .with_dataloader()
-            .with_strategy()
-            .build()
+            TrainContextBuilder(self.train_config).with_checkpoint(checkpoint).build()
         )
 
     def _call_callbacks(self, method_name: str, context: TrainContext):
@@ -57,7 +53,6 @@ class Trainer:
             master_addr=config.master_addr,
             master_port=config.master_port,
             device_type=config.device_type,
-            device_ids=config.device_ids,
             checkpoint=checkpoint,
         )
 
@@ -72,8 +67,9 @@ class Trainer:
                 context.epoch = epoch
                 self._call_callbacks("on_epoch_begin", context)
 
+                accumulation_steps = max(self.train_config.accumulation_steps, 1)
                 for batch in context.dataloader:
-                    if context.iteration % self.train_config.accumulation_steps == 0:
+                    if context.iteration % accumulation_steps == 0:
                         # 2. step
                         self._call_callbacks("on_step_begin", context)
                         context.optimizer.step()
@@ -87,7 +83,7 @@ class Trainer:
                     context.iteration += 1
 
                     # to make the loss normalized by accumulation steps
-                    stand_loss = loss / self.train_config.accumulation_steps
+                    stand_loss = loss / accumulation_steps
                     stand_loss.backward()
 
                     self._call_callbacks("on_batch_end", context)

docker-compose.yml

@@ -0,0 +1,42 @@
+services:
+  server:
+    build: .
+    image: astrai:latest
+    ports:
+      - "8000:8000"
+    volumes:
+      - ./params:/app/params:ro
+      - ./checkpoints:/app/checkpoints
+    command: python -m scripts.tools.server --port 8000 --device cuda
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: 1
+              capabilities: [gpu]
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://localhost:8000/health"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 60s
+    restart: unless-stopped
+
+  server-cpu:
+    profiles: [cpu]
+    build: .
+    image: astrai:latest
+    ports:
+      - "8000:8000"
+    volumes:
+      - ./params:/app/params:ro
+      - ./checkpoints:/app/checkpoints
+    command: python -m scripts.tools.server --port 8000 --device cpu
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://localhost:8000/health"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 120s
+    restart: unless-stopped

scripts/demo/stream_chat.py

@@ -15,7 +15,7 @@ def chat():
     tokenizer = AutoTokenizer.from_pretrained(PARAMETER_ROOT)
     model.to(device="cuda", dtype=torch.bfloat16)
 
-    messages = []
+    messages = [{"role": "system", "content": "You are a helpful assistant."}]
     engine = InferenceEngine(model=model, tokenizer=tokenizer)
 
     while True:

scripts/tools/benchmark.py

@@ -1,9 +1,14 @@
+"""Benchmark Transformer with PagedCache (replaces old persistent_key_values)."""
+
 from dataclasses import dataclass
 from typing import Any, Dict
 
 import torch
+from torch import Tensor
 
-from astrai.model.transformer import ModelConfig, Transformer
+from astrai.config import ModelConfig
+from astrai.inference.cache import PagedCache
+from astrai.model.transformer import Transformer
 
 
 @dataclass
@@ -19,27 +24,25 @@ class GenerationBenchmark:
         self,
         config: ModelConfig,
         device: str = "cuda",
-        dtype: torch.dtype = torch.float16,
+        dtype: torch.dtype = torch.bfloat16,
+        page_size: int = 128,
     ):
         self.config = config
         self.device = device
         self.dtype = dtype
         self.model = Transformer(config).to(device=device, dtype=dtype)
         self.model.eval()
-
-    def _initialize_kv_cache(self, batch_size: int) -> list:
-        """初始化KV缓存"""
-        config = self.config
-        shape = (
-            batch_size,
-            config.max_len,
+        head_dim = config.dim // config.n_heads
+        n_pages = (config.max_len * 4 + page_size - 1) // page_size
+        self._page_cache = PagedCache(
             config.n_layers,
+            n_pages,
+            page_size,
             config.n_kv_heads,
-            config.dim // config.n_heads,
+            head_dim,
+            device,
+            dtype,
         )
-        k_cache = torch.zeros(shape, device=self.device, dtype=self.dtype)
-        v_cache = torch.zeros(shape, device=self.device, dtype=self.dtype)
-        return (k_cache, v_cache)
 
     def _prepare_inputs(self, batch_size: int, prompt_length: int, total_length: int):
         prompt_ids = torch.randint(
@@ -49,7 +52,6 @@ class GenerationBenchmark:
             device=self.device,
             dtype=torch.long,
         )
-
         gen_ids = torch.randint(
             low=0,
             high=self.config.vocab_size,
@@ -57,9 +59,11 @@ class GenerationBenchmark:
             device=self.device,
             dtype=torch.long,
         )
-
         return prompt_ids, gen_ids
 
+    def _make_mask(self, batch_size: int, seq_len: int) -> Tensor:
+        return torch.ones(batch_size, seq_len, dtype=torch.bool, device=self.device)
+
     @torch.inference_mode()
     def run_prefill_benchmark(
         self,
@@ -67,13 +71,11 @@ class GenerationBenchmark:
         prompt_length: int = 512,
         num_trials: int = 10,
     ) -> BenchmarkResult:
-
         for _ in range(3):
             prompt_ids, _ = self._prepare_inputs(
                 batch_size, prompt_length, prompt_length
             )
             _ = self.model(prompt_ids)
-
         torch.cuda.synchronize()
 
         total_time = 0.0
@@ -83,20 +85,20 @@ class GenerationBenchmark:
             prompt_ids, _ = self._prepare_inputs(
                 batch_size, prompt_length, prompt_length
             )
-            start_event = torch.cuda.Event(enable_timing=True)
-            end_event = torch.cuda.Event(enable_timing=True)
+            start = torch.cuda.Event(enable_timing=True)
+            end = torch.cuda.Event(enable_timing=True)
 
-            start_event.record()
+            start.record()
             _ = self.model(prompt_ids)
-            end_event.record()
+            end.record()
             torch.cuda.synchronize()
 
-            trial_time = start_event.elapsed_time(end_event) / 1000
+            trial_time = start.elapsed_time(end) / 1000
             total_time += trial_time
 
             print(
-                f"Trial {trial + 1}/{num_trials}: {prompt_length} tokens in {trial_time:.3f}s "
-                f"({prompt_length / trial_time:.1f} tokens/s)"
+                f"  Trial {trial + 1}/{num_trials}: {prompt_length} tokens in {trial_time:.3f}s "
+                f"({prompt_length / trial_time:.1f} tok/s)"
             )
 
         return BenchmarkResult(
@@ -107,7 +109,7 @@ class GenerationBenchmark:
                 "benchmark_type": "prefill",
                 "batch_size": batch_size,
                 "prompt_length": prompt_length,
-                "dtype": self.dtype,
+                "dtype": str(self.dtype),
                 "device": self.device,
             },
         )
@@ -120,41 +122,62 @@ class GenerationBenchmark:
         gen_length: int = 128,
         num_trials: int = 5,
     ) -> BenchmarkResult:
-
         total_time = 0.0
         total_tokens = batch_size * gen_length * num_trials
+        page_size = self._page_cache.page_size
 
         for trial in range(num_trials):
             prompt_ids, gen_ids = self._prepare_inputs(
-                batch_size, prompt_length, prompt_length + gen_length
+                batch_size,
+                prompt_length,
+                prompt_length + gen_length,
+            )
+
+            n_pages = (prompt_length + gen_length + page_size - 1) // page_size
+            pages = self._page_cache.alloc_n(n_pages * batch_size)
+            page_table = torch.tensor(
+                [pages[i * n_pages : (i + 1) * n_pages] for i in range(batch_size)],
+                dtype=torch.long,
+                device=self.device,
+            )
+
+            cv = self._page_cache.bind(page_table, total_len=prompt_length)
+            _ = self.model(
+                prompt_ids,
+                paged_cache=cv,
+                start_pos=0,
+                input_mask=self._make_mask(batch_size, prompt_length),
             )
-            kv_cache = self._initialize_kv_cache(batch_size)
-            _ = self.model(prompt_ids, persistent_key_values=kv_cache, start_pos=0)
 
             torch.cuda.synchronize()
 
-            start_event = torch.cuda.Event(enable_timing=True)
-            end_event = torch.cuda.Event(enable_timing=True)
-
-            start_event.record()
+            start = torch.cuda.Event(enable_timing=True)
+            end = torch.cuda.Event(enable_timing=True)
 
+            start.record()
             current_pos = prompt_length
             for i in range(gen_length):
                 input_token = gen_ids[:, i : i + 1]
+                cv = self._page_cache.bind(page_table, total_len=current_pos + 1)
                 _ = self.model(
-                    input_token, persistent_key_values=kv_cache, start_pos=current_pos
+                    input_token,
+                    paged_cache=cv,
+                    start_pos=current_pos,
+                    input_mask=self._make_mask(batch_size, 1),
                 )
                 current_pos += 1
-
-            end_event.record()
+            end.record()
             torch.cuda.synchronize()
 
-            trial_time = start_event.elapsed_time(end_event) / 1000
+            trial_time = start.elapsed_time(end) / 1000
             total_time += trial_time
 
+            for idx in pages:
+                self._page_cache.free(idx)
+
             print(
-                f"Trial {trial + 1}/{num_trials}: {gen_length} tokens in {trial_time:.3f}s "
-                f"({gen_length / trial_time:.1f} tokens/s)"
+                f"  Trial {trial + 1}/{num_trials}: {gen_length} tokens in {trial_time:.3f}s "
+                f"({gen_length / trial_time:.1f} tok/s)"
             )
 
         return BenchmarkResult(
@@ -166,31 +189,21 @@ class GenerationBenchmark:
                 "batch_size": batch_size,
                 "prompt_length": prompt_length,
                 "gen_length": gen_length,
-                "dtype": self.dtype,
+                "dtype": str(self.dtype),
                 "device": self.device,
             },
         )
 
 
 def print_benchmark_result(result: BenchmarkResult):
-    """打印基准测试结果"""
-    benchmark_type = result.metadata["benchmark_type"]
-
-    print(f"\n{' ' + benchmark_type.upper().replace('_', ' ') + ' Benchmark ':-^80}")
+    btype = result.metadata["benchmark_type"]
+    print(f"\n{' ' + btype.upper() + ' Benchmark ':-^80}")
     print(f"Total Tokens Processed: {result.total_tokens:,}")
     print(f"Time Consumed: {result.total_time:.3f}s")
-    print(f"Throughput: {result.tokens_per_second:,.1f} tokens/s")
-
-    if benchmark_type == "prefill":
-        print(
-            f"Batch Size: {result.metadata['batch_size']} | Prompt Length: {result.metadata['prompt_length']}"
-        )
-    elif benchmark_type == "decoding":
-        print(
-            f"Batch Size: {result.metadata['batch_size']} | Gen Length: {result.metadata['gen_length']}"
-        )
-
-    print(f"Device: {result.metadata['device']} | Dtype: {result.metadata['dtype']}")
+    print(f"Throughput: {result.tokens_per_second:,.1f} tok/s")
+    for k, v in result.metadata.items():
+        if k != "benchmark_type":
+            print(f"{k.replace('_', ' ').title()}: {v}")
     print("-" * 80)
 
 
@@ -209,15 +222,20 @@ if __name__ == "__main__":
     benchmark = GenerationBenchmark(config)
 
     print("=" * 80)
-    print("Running Transformer Generation Benchmark")
+    print("Running Transformer Generation Benchmark (PagedCache)")
     print("=" * 80)
 
     prefill_result = benchmark.run_prefill_benchmark(
-        batch_size=4, prompt_length=512, num_trials=5
+        batch_size=4,
+        prompt_length=512,
+        num_trials=5,
     )
     print_benchmark_result(prefill_result)
 
     gen_result = benchmark.run_decoding_benchmark(
-        batch_size=4, prompt_length=512, gen_length=128, num_trials=5
+        batch_size=4,
+        prompt_length=512,
+        gen_length=128,
+        num_trials=5,
     )
     print_benchmark_result(gen_result)

scripts/tools/generate.py

@@ -9,7 +9,7 @@ from astrai.tokenize import AutoTokenizer
 
 
 def processor(
-    model_dir: str,
+    param_path: str,
     input_json_file: str,
     output_json_file: str,
     temperature: float,
@@ -20,8 +20,8 @@ def processor(
     max_tokens: int,
 ):
     # Load model and tokenizer
-    model = AutoModel.from_pretrained(model_dir)
-    tokenizer = AutoTokenizer.from_pretrained(model_dir)
+    model = AutoModel.from_pretrained(param_path)
+    tokenizer = AutoTokenizer.from_pretrained(param_path)
     model.to(device="cuda", dtype=torch.bfloat16)
 
     # Create inference engine
@@ -72,7 +72,7 @@ if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="Run generate with a Khaosz model.")
 
     parser.add_argument(
-        "--model_dir", type=str, required=True, help="Path to the model directory."
+        "--param_path", type=str, required=True, help="Path to the model directory."
     )
     parser.add_argument(
         "--input_json_file",

scripts/tools/train.py

@@ -23,7 +23,7 @@ def parse_args() -> argparse.Namespace:
         "--train_type",
         type=str,
         required=True,
-        choices=["seq", "sft", "dpo"],
+        choices=["seq", "sft", "dpo", "grpo"],
         help="Train type.",
     )
     parser.add_argument(
@@ -42,9 +42,7 @@ def parse_args() -> argparse.Namespace:
     parser.add_argument(
         "--n_epoch", type=int, default=1, help="Number of epochs to train."
     )
-    parser.add_argument(
-        "--batch_size", type=int, default=1, help="Batch size for training."
-    )
+    parser.add_argument("--batch_size", type=int, default=1, help="Batch size per GPU.")
     parser.add_argument(
         "--accumulation_steps",
         type=int,
@@ -55,7 +53,7 @@ def parse_args() -> argparse.Namespace:
         "--warmup_steps",
         type=int,
         default=1000,
-        help="Number of iters between warnings.",
+        help="Number of warmup steps for LR scheduler.",
     )
     parser.add_argument(
         "--max_lr", type=float, default=3e-4, help="Max learning rate for training."
@@ -100,12 +98,19 @@ def parse_args() -> argparse.Namespace:
         "--window_size",
         type=int,
         default=None,
-        help="the max length of the input sequence.",
+        help="Max length of the input sequence.",
     )
     parser.add_argument(
-        "--stride", type=int, default=None, help="the step size of the input sequence."
+        "--stride", type=int, default=None, help="Step size of the input sequence."
     )
     parser.add_argument("--dpo_beta", type=float, default=0.1, help="DPO beta value.")
+    parser.add_argument("--group_size", type=int, default=4, help="GRPO group size.")
+    parser.add_argument(
+        "--grpo_clip_eps", type=float, default=0.2, help="GRPO clipping epsilon."
+    )
+    parser.add_argument(
+        "--grpo_kl_coef", type=float, default=0.01, help="GRPO KL penalty coefficient."
+    )
     parser.add_argument(
         "--label_smoothing",
         type=float,
@@ -125,6 +130,12 @@ def parse_args() -> argparse.Namespace:
         default="checkpoint",
         help="Directory to save checkpoints.",
     )
+    parser.add_argument(
+        "--grpo_sync_interval",
+        type=int,
+        default=200,
+        help="GRPO ref model sync interval (steps).",
+    )
     parser.add_argument(
         "--start_epoch", type=int, default=0, help="Start epoch for training."
     )
@@ -144,7 +155,7 @@ def parse_args() -> argparse.Namespace:
 
 def ddp_wrap(model: nn.Module):
     local_rank = get_rank()
-    model = model.to(device=f"cuda:{local_rank}", dtype=torch.bfloat16)
+    model = model.to(dtype=torch.bfloat16)
     ddp_model = DDP(
         model,
         device_ids=[local_rank],
@@ -182,6 +193,10 @@ def train(
     ckpt_interval: int,
     ckpt_dir: str,
     dpo_beta: float,
+    grpo_clip_eps: float,
+    grpo_kl_coef: float,
+    group_size: int,
+    grpo_sync_interval: int,
     adamw_beta1: float,
     adamw_beta2: float,
     adamw_weight_decay: float,
@@ -195,7 +210,7 @@ def train(
     nprocs: int,
     device_type: str,
 ):
-    assert train_type in ["seq", "sft", "dpo"]
+    assert train_type in ["seq", "sft", "dpo", "grpo"]
     assert os.path.exists(param_path)
 
     # Load config
@@ -216,7 +231,14 @@ def train(
         state_dict = st.load_file(weights_path)
         model.load_state_dict(state_dict, strict=False)
 
-    strategy_kwargs = {"dpo_beta": dpo_beta, "label_smoothing": label_smoothing}
+    strategy_kwargs = {
+        "dpo_beta": dpo_beta,
+        "label_smoothing": label_smoothing,
+        "clip_eps": grpo_clip_eps,
+        "kl_coef": grpo_kl_coef,
+        "group_size": group_size,
+        "sync_interval": grpo_sync_interval,
+    }
 
     dataset = DatasetFactory.load(
         train_type=train_type,

tests/inference/conftest.py

@@ -14,37 +14,32 @@ def client():
     return TestClient(app)
 
 
-@pytest.fixture
-def mock_model_param():
-    """Create a mock ModelParameter."""
-    mock_param = MagicMock()
-    mock_param.model = MagicMock()
-    mock_param.tokenizer = MagicMock()
-    mock_param.config = MagicMock()
-    mock_param.config.max_len = 100
-    mock_param.tokenizer.encode = MagicMock(return_value=[1, 2, 3])
-    mock_param.tokenizer.decode = MagicMock(return_value="mock response")
-    mock_param.tokenizer.stop_ids = []
-    mock_param.tokenizer.pad_id = 0
-    return mock_param
-
-
 @pytest.fixture
 def mock_engine():
     """Create a mock InferenceEngine."""
+
+    async def _async_gen():
+        yield "chunk1"
+        yield "chunk2"
+        yield "[DONE]"
+
     mock = MagicMock()
     mock.generate.return_value = "mock response"
+    mock.generate_async.return_value = _async_gen()
     mock.get_stats.return_value = {
         "total_tasks": 0,
         "total_tokens": 0,
         "active_tasks": 0,
         "waiting_queue": 0,
     }
+    mock.tokenizer.encode.return_value = [1, 2, 3]
+    mock.tokenizer.decode.return_value = "mock response"
+    mock.tokenizer.apply_chat_template.return_value = "mock prompt"
     return mock
 
 
 @pytest.fixture
-def loaded_model(mock_model_param, monkeypatch):
-    """Simulate that the model is loaded."""
-    monkeypatch.setattr("astrai.inference.server._model_param", mock_model_param)
-    return mock_model_param
+def loaded_model(mock_engine, monkeypatch):
+    """Simulate that the engine is loaded."""
+    monkeypatch.setattr("astrai.inference.server._state.engine", mock_engine)
+    return mock_engine

tests/inference/test_scheduler_concurrency.py

@@ -5,103 +5,9 @@ import time
 from unittest.mock import MagicMock, patch
 
 import pytest
+import torch
 
-from astrai.inference.scheduler import (
-    InferenceScheduler,
-    PrefixCacheManager,
-)
-
-
-def test_prefix_cache_concurrent_insert_find():
-    """Test concurrent insert and find operations."""
-    cache = PrefixCacheManager(max_capacity=100)
-
-    results = {"errors": [], "inserts": 0, "finds": 0}
-
-    def insert_worker():
-        try:
-            for i in range(50):
-                cache.insert((i,), slot=i % 10)
-                results["inserts"] += 1
-        except Exception as e:
-            results["errors"].append(str(e))
-
-    def find_worker():
-        try:
-            for i in range(50):
-                cache.find_longest_prefix([i])
-                results["finds"] += 1
-        except Exception as e:
-            results["errors"].append(str(e))
-
-    threads = [threading.Thread(target=insert_worker) for _ in range(3)]
-    threads += [threading.Thread(target=find_worker) for _ in range(3)]
-
-    for t in threads:
-        t.start()
-    for t in threads:
-        t.join()
-
-    assert len(results["errors"]) == 0, f"Errors: {results['errors']}"
-    assert results["inserts"] == 150
-    assert results["finds"] == 150
-
-
-def test_prefix_cache_concurrent_release():
-    """Test concurrent release operations."""
-    cache = PrefixCacheManager(max_capacity=100)
-
-    # Insert some prefixes
-    for i in range(10):
-        cache.insert((i,), slot=i)
-
-    results = {"errors": []}
-
-    def release_worker():
-        try:
-            for i in range(10):
-                cache.release((i,))
-        except Exception as e:
-            results["errors"].append(str(e))
-
-    threads = [threading.Thread(target=release_worker) for _ in range(3)]
-
-    for t in threads:
-        t.start()
-    for t in threads:
-        t.join()
-
-    assert len(results["errors"]) == 0, f"Errors: {results['errors']}"
-
-
-def test_prefix_cache_concurrent_insert_release_find():
-    """Test mixed concurrent operations."""
-    cache = PrefixCacheManager(max_capacity=50)
-
-    results = {"errors": []}
-
-    def worker(worker_id):
-        try:
-            for i in range(20):
-                token_ids = (worker_id * 100 + i,)
-                cache.insert(token_ids, slot=worker_id)
-
-                # Find after insert
-                cache.find_longest_prefix(list(token_ids))
-
-                # Release
-                cache.release(token_ids)
-        except Exception as e:
-            results["errors"].append(f"Worker {worker_id}: {str(e)}")
-
-    threads = [threading.Thread(target=worker, args=(i,)) for i in range(5)]
-
-    for t in threads:
-        t.start()
-    for t in threads:
-        t.join()
-
-    assert len(results["errors"]) == 0, f"Errors: {results['errors']}"
+from astrai.inference.scheduler import InferenceScheduler
 
 
 @pytest.fixture
@@ -114,6 +20,9 @@ def mock_model_and_tokenizer():
     mock_model.config.dim = 128
     mock_model.config.n_layers = 2
     mock_model.config.max_len = 100
+    mock_model.parameters.return_value = iter(
+        [MagicMock(dtype=torch.float32, device=torch.device("cpu"))]
+    )
 
     mock_tokenizer = MagicMock()
     mock_tokenizer.encode.return_value = [1, 2, 3, 4, 5]
@@ -266,55 +175,3 @@ def test_scheduler_concurrent_get_stats(mock_model_and_tokenizer):
     for stats in results["stats"]:
         assert "total_tasks" in stats
         assert stats["total_tasks"] >= 0
-
-
-def test_prefix_cache_insert_same_prefix_concurrently():
-    """Test inserting the same prefix concurrently."""
-    cache = PrefixCacheManager(max_capacity=100)
-
-    results = {"slot_values": [], "errors": []}
-
-    def insert_worker():
-        try:
-            # All workers try to insert the same prefix
-            cache.insert((1, 2, 3), slot=threading.current_thread().name)
-            node = cache.root.children.get(1)
-            if node:
-                node = node.children.get(2)
-                if node:
-                    node = node.children.get(3)
-                    if node:
-                        results["slot_values"].append(node.slot)
-        except Exception as e:
-            results["errors"].append(str(e))
-
-    threads = [threading.Thread(target=insert_worker) for _ in range(10)]
-
-    for t in threads:
-        t.start()
-    for t in threads:
-        t.join()
-
-    # All inserts should succeed, final slot should be one of the values
-    assert len(results["errors"]) == 0, f"Errors: {results['errors']}"
-    # Check ref_count is correct (should be 10)
-    node = cache.root.children.get(1).children.get(2).children.get(3)
-    assert node.ref_count == 10, f"Expected ref_count=10, got {node.ref_count}"
-
-
-def test_prefix_cache_ref_count_underflow_prevention():
-    """Test that ref_count doesn't go negative."""
-    cache = PrefixCacheManager(max_capacity=100)
-
-    # Insert a prefix
-    cache.insert((1, 2, 3), slot=0)
-
-    # Release multiple times
-    for _ in range(5):
-        cache.release((1, 2, 3))
-
-    # Try to find it - should return None since ref_count would be negative
-    # or handle it gracefully
-    node = cache.root.children.get(1).children.get(2).children.get(3)
-    # The ref_count should be 0, not negative
-    assert node.ref_count >= 0, f"ref_count went negative: {node.ref_count}"

tests/inference/test_server.py

@@ -4,88 +4,38 @@ import pytest
 
 
 def test_health_no_model(client, monkeypatch):
-    """GET /health should return 200 even when model not loaded."""
-    monkeypatch.setattr("astrai.inference.server._model_param", None)
-    monkeypatch.setattr("astrai.inference.server._engine", None)
+    """GET /health should return 200 even when engine not loaded."""
+    monkeypatch.setattr("astrai.inference.server._state.engine", None)
     response = client.get("/health")
     assert response.status_code == 200
     data = response.json()
     assert data["status"] == "ok"
     assert not data["model_loaded"]
-    assert not data["engine_ready"]
 
 
-def test_health_with_model(client, loaded_model, mock_engine, monkeypatch):
-    """GET /health should return 200 when model is loaded."""
-    monkeypatch.setattr("astrai.inference.server._engine", mock_engine)
+def test_health_with_model(client, loaded_model):
+    """GET /health should return 200 when engine is loaded."""
     response = client.get("/health")
     assert response.status_code == 200
     data = response.json()
     assert data["status"] == "ok"
     assert data["model_loaded"] is True
-    assert data["engine_ready"] is True
 
 
-def test_generate_non_stream(client, loaded_model, mock_engine, monkeypatch):
-    """POST /generate with stream=false should return JSON response."""
-    monkeypatch.setattr("astrai.inference.server._engine", mock_engine)
-    response = client.post(
-        "/generate",
-        params={
-            "query": "Hello",
-            "temperature": 0.8,
-            "top_p": 0.95,
-            "top_k": 50,
-            "max_len": 100,
-            "stream": False,
-        },
-    )
-    assert response.status_code == 200
-    data = response.json()
-    assert data["response"] == "mock response"
+def test_chat_completions_non_stream(client, loaded_model, monkeypatch):
+    """POST /v1/chat/completions with stream=false returns OpenAI-style JSON."""
 
+    async def async_gen():
+        yield "Assistant reply"
 
-def test_generate_stream(client, loaded_model, mock_engine, monkeypatch):
-    """POST /generate with stream=true should return plain text stream."""
-
-    # Create a streaming mock
-    def stream_gen():
-        yield "chunk1"
-        yield "chunk2"
-
-    mock_engine.generate.return_value = stream_gen()
-    monkeypatch.setattr("astrai.inference.server._engine", mock_engine)
-    response = client.post(
-        "/generate",
-        params={
-            "query": "Hello",
-            "temperature": 0.8,
-            "top_p": 0.95,
-            "top_k": 50,
-            "max_len": 100,
-            "stream": True,
-        },
-        headers={"Accept": "text/plain"},
-    )
-    assert response.status_code == 200
-    assert response.headers["content-type"] == "text/plain; charset=utf-8"
-    # The stream yields lines ending with newline
-    content = response.content.decode("utf-8")
-    assert "chunk1" in content
-    assert "chunk2" in content
-
-
-def test_chat_completions_non_stream(client, loaded_model, mock_engine, monkeypatch):
-    """POST /v1/chat/completions with stream=false returns OpenAI‑style JSON."""
-    mock_engine.generate.return_value = "Assistant reply"
-    monkeypatch.setattr("astrai.inference.server._engine", mock_engine)
+    mock_engine = loaded_model
+    mock_engine.generate_async.return_value = async_gen()
+    monkeypatch.setattr("astrai.inference.server._state.engine", mock_engine)
     response = client.post(
         "/v1/chat/completions",
         json={
             "messages": [{"role": "user", "content": "Hello"}],
             "temperature": 0.8,
-            "top_p": 0.95,
-            "top_k": 50,
             "max_tokens": 100,
             "stream": False,
         },
@@ -94,57 +44,120 @@ def test_chat_completions_non_stream(client, loaded_model, mock_engine, monkeypa
     data = response.json()
     assert data["object"] == "chat.completion"
     assert len(data["choices"]) == 1
-    assert data["choices"][0]["message"]["content"] == "Assistant reply"
+    assert "usage" in data
+    assert "prompt_tokens" in data["usage"]
 
 
-def test_chat_completions_stream(client, loaded_model, mock_engine, monkeypatch):
+def test_chat_completions_stream(client, loaded_model, monkeypatch):
     """POST /v1/chat/completions with stream=true returns SSE stream."""
 
-    # Simulate a streaming generator that yields cumulative responses
-    def stream_gen():
+    async def async_gen():
         yield "cumulative1"
         yield "cumulative2"
-        yield "[DONE]"
 
-    mock_engine.generate.return_value = stream_gen()
-    monkeypatch.setattr("astrai.inference.server._engine", mock_engine)
+    mock_engine = loaded_model
+    mock_engine.generate_async.return_value = async_gen()
+    monkeypatch.setattr("astrai.inference.server._state.engine", mock_engine)
     response = client.post(
         "/v1/chat/completions",
         json={
             "messages": [{"role": "user", "content": "Hello"}],
             "temperature": 0.8,
-            "top_p": 0.95,
-            "top_k": 50,
             "max_tokens": 100,
             "stream": True,
         },
         headers={"Accept": "text/event-stream"},
     )
     assert response.status_code == 200
-    assert response.headers["content-type"] == "text/event-stream; charset=utf-8"
-    # Parse SSE lines
     lines = [
         line.strip() for line in response.content.decode("utf-8").split("\n") if line
     ]
-    # Should contain data lines and a final [DONE]
     assert any("cumulative1" in line for line in lines)
     assert any("cumulative2" in line for line in lines)
+    assert any("[DONE]" in line for line in lines)
 
 
-def test_generate_with_history(client, loaded_model, mock_engine, monkeypatch):
-    """POST /generate with history parameter."""
-    monkeypatch.setattr("astrai.inference.server._engine", mock_engine)
+def test_messages_non_stream(client, loaded_model, monkeypatch):
+    """POST /v1/messages with stream=false returns Anthropic-style JSON."""
+
+    async def async_gen():
+        yield "Assistant reply"
+
+    mock_engine = loaded_model
+    mock_engine.generate_async.return_value = async_gen()
+    monkeypatch.setattr("astrai.inference.server._state.engine", mock_engine)
     response = client.post(
-        "/generate",
-        params={
-            "query": "Hi",
-            "history": [["user1", "assistant1"], ["user2", "assistant2"]],
+        "/v1/messages",
+        json={
+            "messages": [{"role": "user", "content": "Hello"}],
+            "temperature": 0.8,
+            "max_tokens": 100,
             "stream": False,
         },
     )
     assert response.status_code == 200
-    # Verify the engine.generate was called
-    mock_engine.generate.assert_called_once()
+    data = response.json()
+    assert data["type"] == "message"
+    assert data["role"] == "assistant"
+    assert len(data["content"]) == 1
+    assert data["content"][0]["type"] == "text"
+    assert "usage" in data
+    assert "input_tokens" in data["usage"]
+
+
+def test_messages_stream(client, loaded_model, monkeypatch):
+    """POST /v1/messages with stream=true returns Anthropic SSE stream."""
+
+    async def async_gen():
+        yield "cumulative1"
+        yield "cumulative2"
+
+    mock_engine = loaded_model
+    mock_engine.generate_async.return_value = async_gen()
+    monkeypatch.setattr("astrai.inference.server._state.engine", mock_engine)
+    response = client.post(
+        "/v1/messages",
+        json={
+            "messages": [{"role": "user", "content": "Hello"}],
+            "temperature": 0.8,
+            "max_tokens": 100,
+            "stream": True,
+        },
+        headers={"Accept": "text/event-stream"},
+    )
+    assert response.status_code == 200
+    content = response.content.decode("utf-8")
+    assert "message_start" in content
+    assert "content_block_start" in content
+    assert "content_block_delta" in content
+    assert "cumulative1" in content
+    assert "cumulative2" in content
+    assert "content_block_stop" in content
+    assert "message_delta" in content
+    assert "message_stop" in content
+
+
+def test_messages_with_system(client, loaded_model, monkeypatch):
+    """POST /v1/messages with system prompt."""
+
+    async def async_gen():
+        yield "Reply"
+
+    mock_engine = loaded_model
+    mock_engine.generate_async.return_value = async_gen()
+    monkeypatch.setattr("astrai.inference.server._state.engine", mock_engine)
+    response = client.post(
+        "/v1/messages",
+        json={
+            "messages": [{"role": "user", "content": "Hello"}],
+            "system": "You are a helpful assistant.",
+            "max_tokens": 100,
+            "stream": False,
+        },
+    )
+    assert response.status_code == 200
+    data = response.json()
+    assert data["type"] == "message"
 
 
 if __name__ == "__main__":

tests/trainer/test_train_strategy.py

@@ -72,6 +72,7 @@ def test_schedule_factory_random_configs():
 
             # Test scheduler step functionality
             initial_lr = scheduler.get_last_lr()
+            optimizer.step()
             scheduler.step()
             new_lr = scheduler.get_last_lr()
 
@@ -112,6 +113,7 @@ def test_schedule_factory_edge_cases():
 
         # Test multiple steps
         for _ in range(10):
+            optimizer.step()
             scheduler.step()
 
 
@@ -136,6 +138,7 @@ def test_schedule_factory_state_persistence():
 
     # Take a few steps
     for _ in range(5):
+        optimizer.step()
         scheduler.step()
 
     # Save state