ViperEkura a29bdfae46 refactor: rework attention backend resolution ...

- explicit attn_backend() context wins over ASTR_BACKEND env
- polymorphic available()/supports_call() replace isinstance dispatch
- cache singleton backend instances to avoid hot-path allocation
- training (fwd=None) resolves cuda > flash > torch by capability
- flash dense supports mask-free calls only; masked training falls back to torch

2026-08-23 14:47:02 +08:00

..

config

refactor: simplify training and inference interfaces

2026-08-19 20:55:13 +08:00

dataset

refactor: deduplicate low-risk code paths

2026-08-19 16:17:40 +08:00

extension

refactor: rework attention backend resolution

2026-08-23 14:47:02 +08:00

inference

refactor: rework attention backend resolution

2026-08-23 14:47:02 +08:00

model

feat: load HuggingFace checkpoints via key/config conversion

2026-08-20 11:34:59 +08:00

optim

fix: use torch.optim.AdamW in ManoAdamW instead of NAdamW

2026-08-01 09:20:54 +08:00

parallel

refactor: deduplicate low-risk code paths

2026-08-19 16:17:40 +08:00

preprocessing

fix : correct prefill sampling and record alignment

2026-08-05 22:20:29 +08:00

serialization

refactor: unify CUDA skip guards in tests

2026-08-22 21:08:06 +08:00

tokenize

refactor: simplify BaseFactory and separate ModelFactory from AutoModel

2026-07-30 09:38:20 +08:00

trainer

feat: load HuggingFace checkpoints via key/config conversion

2026-08-20 11:34:59 +08:00

__init__.py

refactor: simplify training and inference interfaces

2026-08-19 20:55:13 +08:00

factory.py

refactor: simplify BaseFactory and separate ModelFactory from AutoModel

2026-07-30 09:38:20 +08:00

logging.py

feat: add distributed rank to logs

2026-08-20 01:37:00 +08:00

protocols.py

refactor: 重构训练后端为 Executor 模式

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

signal_handler.py

refactor: move signal_handler from parallel/ to top-level for broader reuse

2026-07-28 10:36:17 +08:00