DefTruth e9528f6dc6

[Kernel] support merge_attn_states CUDA kernel, 3x speedup (#16173) ...

Signed-off-by: DefTruth <qiustudent_r@163.com>

2025-04-11 06:50:50 -06:00

..

attention

[Kernel] support merge_attn_states CUDA kernel, 3x speedup (#16173)

2025-04-11 06:50:50 -06:00

core

Enforce valid max_num_batched_tokens when disable_chunked_mm_input=True (#16447)

2025-04-11 08:09:52 +00:00

engine

[V1] Zero-copy tensor/ndarray serialization/transmission (#13790)

2025-04-10 19:23:14 +00:00

executor

[BugFix] fix some typos found by typos. (#16314)

2025-04-09 03:43:59 -07:00

metrics

[Metrics] Add bucket for request_latency, time_to_first_token and time_per_output_token (#15202)

2025-04-06 20:34:51 -07:00

sample

[TPU][V1] Disable per-request seed/Generator (#16172)

2025-04-10 17:05:44 -04:00

spec_decode

[V1][Spec Decode] Eagle Model loading (#16035)

2025-04-10 11:21:48 -07:00

stats

Update deprecated Python 3.8 typing (#13971)

2025-03-02 17:34:51 -08:00

structured_output

[Core] Upgrade to xgrammar 0.1.18, add cache size limit (#16283)

2025-04-08 19:13:22 -07:00

worker

[TPU][V1] Disable per-request seed/Generator (#16172)

2025-04-10 17:05:44 -04:00

__init__.py

[V1] AsyncLLM Implementation (#9826)

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

kv_cache_interface.py

[V1] Implement sliding window attention in kv_cache_manager (#14097)

2025-04-01 00:33:17 -07:00

outputs.py

[V1] Aggregate chunked prompt logprobs in model runner (#14875)

2025-03-24 12:27:57 -04:00

request.py

[Bugfix] Avoid transferring cached multi-modal items from P0 to P1 (#16273)

2025-04-09 00:51:27 -07:00

serial_utils.py

[V1] Zero-copy tensor/ndarray serialization/transmission (#13790)

2025-04-10 19:23:14 +00:00

utils.py

[V1][BugFix] Exit properly if engine core fails during startup (#16137)

2025-04-07 15:30:15 -07:00