Wang Yijun 44554a0068

Add tokenization_kwargs to encode for embedding model truncation (#21033)

2025-07-22 08:24:00 -07:00

..

attention

[Attention] Clean up iRoPE in V1 (#21188)

2025-07-21 09:10:30 -07:00

core

[feat] Enable mm caching for transformers backend (#21358)

2025-07-22 08:18:46 -07:00

engine

Add tokenization_kwargs to encode for embedding model truncation (#21033)

2025-07-22 08:24:00 -07:00

executor

[BugFix] Make PD work with Ray (#21072)

2025-07-19 08:46:50 -07:00

metrics

[DP] Fix Prometheus Logging (#21257)

2025-07-21 09:11:35 -07:00

pool

[Model][1/N] Support multiple poolers at model level (#21227)

2025-07-21 02:22:21 -07:00

sample

[Core] Optimize update checks in LogitsProcessor (#21245)

2025-07-22 05:27:18 -07:00

spec_decode

[Attention] Refactor attention metadata builder interface (#20466)

2025-07-17 04:44:25 +00:00

structured_output

[Refactor][V1] Move outlines utils for V1 imports (#20878)

2025-07-14 00:58:35 -07:00

worker

[XPU] Enable external_launcher to serve as an executor via torchrun (#21021)

2025-07-21 21:47:35 -07:00

__init__.py

[V1] AsyncLLM Implementation (#9826)

2024-11-11 23:05:38 +00:00

kv_cache_interface.py

[Core] Support Local Chunked Attention for Hybrid KV Cache (#19351)

2025-07-18 20:48:38 -07:00

outputs.py

Export NaNs in logits to scheduler_stats if output is corrupted (#18777)

2025-06-20 22:47:16 +08:00

request.py

Implement Async Scheduling (#19970)

2025-07-14 23:01:46 -07:00

serial_utils.py

[Core] Do not copy array during hashing (#19484)

2025-06-18 15:36:55 -07:00

utils.py

[MISC] Move bind_kv_cache to worker module (#20900)

2025-07-14 09:40:00 +00:00