ViperEkura f3eaaef842 refactor: remove redundant strategy/executor code ...

- Drop BaseStrategy.model_fn (stored but never read)
- Drop model_fn= passed to StrategyFactory.create in train_context
- Simplify FSDPExecutor.clip_grad_norm None branch to delegate to super()
- Remove DDPExecutor._gather_state_dict override (identical to base)

2026-07-19 12:45:58 +08:00

..

config

feat: make max_grad_norm optional (None disables clipping)

2026-07-19 00:08:18 +08:00

dataset

fix: apply chat template in DPO tokenization

2026-07-19 00:00:51 +08:00

extension

refactor: stride-based attn interface with layout and causal mask

2026-07-14 21:34:42 +08:00

inference

fix: incremental decode to avoid U+FFFD in streaming

2026-07-18 13:05:36 +08:00

model

perf: replace paged KV cache with contiguous ContiguousCache, decode all groups

2026-07-05 11:34:36 +08:00

parallel

refactor: remove redundant strategy/executor code

2026-07-19 12:45:58 +08:00

preprocessing

refactor: deduplicate preprocessing kernel and BFD packing

2026-07-19 12:27:56 +08:00

serialization

feat: add record-mode to Store for DPO/GRPO

2026-07-18 21:02:29 +08:00

tokenize

fix: make tokenizer picklable for spawn multiprocessing

2026-07-19 11:59:55 +08:00

trainer

refactor: remove redundant strategy/executor code

2026-07-19 12:45:58 +08:00

__init__.py

feat: add record-mode to Store for DPO/GRPO

2026-07-18 21:02:29 +08:00

factory.py

fix: ForwardRef._evaluate Python 3.12 compatibility

2026-07-03 18:41:19 +08:00

protocols.py

refactor: 重构训练后端为 Executor 模式

2026-05-24 20:35:44 +08:00