ViperEkura 5416c2e8fb perf: replace paged KV cache with contiguous ContiguousCache, decode all groups ...

- Add KVCache/CacheView abstract base classes in cache.py
- Add ContiguousCache (contiguous per-slot buffer, default) alongside PageCache (paged, renamed from old KVCache)
- Merge make_table_tensor + bind into bind_tasks on KVCache interface
- Remove task_cached/task_record_hashes from base class (PageCache-only)
- Scheduler: decode all position groups instead of just the largest (eliminates 63% group skip rate)
- Scheduler: accept optional cache param for swapping implementations
- Model layer type hints use CacheView base class
- Batch 1-32: 1-7% speedup from eliminating Storage.gather overhead
- All 183 inference tests pass

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

..

components

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

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

__init__.py

feat: 新增LoRA微调模块

2026-05-25 20:15:31 +08:00

automodel.py

refactor : 移除 -> None 返回值标注，拆分 FSDP 参数，新增 mmap 数据集存储

2026-05-28 13:57:06 +08:00

encoder.py

refactor : neftune_alpha 在 Embedding 构造时传入，由模型配置链路负责

2026-06-19 14:23:27 +08:00

transformer.py

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

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