[bench] Support common prefix len config (for decode-only bench) (#29934)

Ming Yang committed 11 months ago

[BugFix] Adding env variable to disable async grammar compilation (#29996)

Alec S committed 11 months ago

[Frontend] Remove deprecated -O.xx flag (#29991)

Yanan Cao committed 11 months ago

[CI/Build][AMD] Use float16 in test_reset_prefix_cache_e2e to avoid accuracy issues (#29997)

rasmith committed 11 months ago

[Frontend][Model] Add 'float16' to possible mamba cache dtype values, override mamba SSM cache dtype value for NemotronH (#29978)

amitz-nv committed 11 months ago

[DOC]: Add kthena to integrations (#29931)

Tiger Xu / Zhonghu Xu committed 11 months ago

[AMD][CI] Add ray[default] Dependency On ROCm To Pass v1/metrics/test_engine_logger_apis.py (#30110)

Micah Williamson committed 11 months ago

Refactor example prompts fixture (#29854)

Chukwuma Nwaugha committed 11 months ago

[Bugfix] Correct num_q_heads on DCP for Flashinfer backends (#29487)

Jingchun Gao committed 11 months ago

[ROCm][CI] Increase the memory threshold for test_deep_sleep_fp8_kvcache (#30104)

Charlie Fu committed 11 months ago

Do not guard during noop elimination pass (#30095)

Laith Sakka committed 11 months ago

[ROCm][CI] Add jiwer dependency for testing (#30081)

Charlie Fu committed 11 months ago

[PCP&DCP] move CUDAGraph check for PCP&DCP to the check func of platforms (#29952)

Qiu committed 11 months ago

[Model] Add Holo2 reasoning parser (#30048)

Hubert de La Jonquiere committed 11 months ago

[CI] fix silent error in nightly wheel index generation script, add generation time to HTML index (#30060)

Shengqi Chen committed 11 months ago

[CI/Build] Update batch invariant test trigger (#30080)

Zhewen Li committed 11 months ago

[CI/Build][AMD] Add Llama4 Maverick FP8 to AMD CI (#28695)

Zhewen Li committed 11 months ago

[Perf] Enable separate shared_experts stream only for CUDA (#30085)

Alexander Matveev committed 11 months ago

docs: update metrics design doc to use new vllm:kv_cache_usage_perc (#30041)

TimWang committed 11 months ago

[aot_compile]change VLLM backend to read fake args from example_value (#29104)

Laith Sakka committed 11 months ago