refactor : align config field names with Hugging Face

- dim -> hidden_size, n_layers -> num_hidden_layers
- dim_ffn -> intermediate_size, n_heads -> num_attention_heads
- n_kv_heads -> num_key_value_heads, max_len -> max_position_embeddings
- norm_eps -> rms_norm_eps, tie_weight -> tie_word_embeddings
- update model, inference, training, scripts, tests, docs

assets/docs/architecture.md

@@ -28,17 +28,17 @@ classDiagram
 
         class AutoRegressiveLMConfig {
             +Optional[int] vocab_size
-            +Optional[int] dim
-            +Optional[int] n_layers
-            +Optional[float] norm_eps
-            +Optional[int] dim_ffn
-            +Optional[bool] tie_weight
+            +Optional[int] hidden_size
+            +Optional[int] num_hidden_layers
+            +Optional[float] rms_norm_eps
+            +Optional[int] intermediate_size
+            +Optional[bool] tie_word_embeddings
             +Optional[dict] rope_scaling
-            +Optional[int] max_len
+            +Optional[int] max_position_embeddings
             +Optional[float] rope_theta
             +str attn_type
-            +Optional[int] n_heads
-            +Optional[int] n_kv_heads
+            +Optional[int] num_attention_heads
+            +Optional[int] num_key_value_heads
             +Optional[bool] use_qk_norm
             +Optional[bool] use_gated_attention
             +Optional[int] kv_lora_rank
@@ -53,15 +53,15 @@ classDiagram
 
         class EncoderConfig {
             +Optional[int] vocab_size
-            +Optional[int] dim
-            +Optional[int] n_layers
-            +Optional[float] norm_eps
-            +Optional[int] dim_ffn
-            +Optional[int] max_len
+            +Optional[int] hidden_size
+            +Optional[int] num_hidden_layers
+            +Optional[float] rms_norm_eps
+            +Optional[int] intermediate_size
+            +Optional[int] max_position_embeddings
             +Optional[float] rope_theta
             +str attn_type
-            +Optional[int] n_heads
-            +Optional[int] n_kv_heads
+            +Optional[int] num_attention_heads
+            +Optional[int] num_key_value_heads
             +Optional[bool] use_qk_norm
             +str ffn_type
             +Optional[dict] rope_scaling

assets/docs/dataflow.md

@@ -85,7 +85,7 @@ All backends normalise tensors into `Store._data[Dict[str, List[Tensor]]]` + `St
 ```
 DatasetFactory.load(train_type, load_path, window_size, stride=None,
                     storage_type=None, tokenizer_path=None,
-                    max_len=2048, store=None)
+                    max_position_embeddings=2048, store=None)
   → BaseDataset.load(load_path, storage_type=None)
     → detect_format(load_path)
     → StoreFactory.create(storage_type)

assets/docs/inference.md

@@ -32,7 +32,7 @@ ContiguousCache (simple contiguous per-slot cache)
   ├── ContiguousCacheView  bundles k/v tensors + slot indices for attention layers
 ```
 
-Created by default when no cache is passed to `InferenceScheduler`. Each task occupies a fixed slot of `[max_seq_len, n_kv_heads, head_dim]`. Simple and efficient for small-to-medium batch sizes.
+Created by default when no cache is passed to `InferenceScheduler`. Each task occupies a fixed slot of `[max_seq_len, num_key_value_heads, head_dim]`. Simple and efficient for small-to-medium batch sizes.
 
 ### PageCache (paged with prefix sharing)
 
@@ -42,7 +42,7 @@ PageCache (paged KV cache with prefix sharing, alternative)
   │     ├── Allocator         bitmask-based page allocator + ref-count + LRU
   │     └── PrefixCache       hash-based prefix matching (page_hash via polynomial hash)
   ├── TaskTable              maps task_id → page_table + cached token count
-  ├── Storage                k_cache / v_cache tensors (n_layers × n_pages × page_size × n_kv_heads × head_dim)
+  ├── Storage                k_cache / v_cache tensors (num_hidden_layers × n_pages × page_size × num_key_value_heads × head_dim)
   └── PageCacheView          bundles Storage + page_table + total_len for attention layers
 ```
 

assets/docs/params.md

@@ -44,7 +44,7 @@ Combined optimizer: matrix parameters via **Muon**, non-matrix via **AdamW** (`f
 
 | Parameter | Description | Default |
 |-----------|-------------|---------|
-| `--window_size` | Max input sequence length | model config `max_len` |
+| `--window_size` | Max input sequence length | model config `max_position_embeddings` |
 | `--stride` | Stride for sliding window over sequences | None |
 | `--random_seed` | Random seed for reproducibility | 3407 |
 | `--num_workers` | DataLoader worker processes | 4 |
@@ -186,7 +186,7 @@ See [Inference Guide](inference.md) for HTTP API documentation.
 | `--top_k` | int | `30` | Top-k filtering |
 | `--top_p` | float | `0.95` | Nucleus sampling threshold |
 | `--batch_size` | int | `1` | Batch size for generation |
-| `--max_tokens` | int | model config `max_len` | Maximum tokens to generate |
+| `--max_tokens` | int | model config `max_position_embeddings` | Maximum tokens to generate |
 
 Usage:
 ```bash