vllm-project/vllm

Pull Requests Commits

[BugFix] Set default random seed to 0 for V1 (#17929)

WoosukKwon committed 276 days ago

Verified 2ff297dc

[Bugfix][V1] Only get input embeddings w/ multi-modal models if first PP (#17916)

jinhuang12 committed 276 days ago

Verified 8dd0671b

[v1][KVCacheManager] Avoid full cache hit by controlling max_length (#17999)

heheda12345 committed 276 days ago

Verified f0d610a8

Fix Broken macro for cutlass moe (#18049)

drisspg committed 276 days ago

Verified e57e4d6e

[BugFix] Fix 4-GPU RLHF tests (#18007)

njhill committed 276 days ago

Verified ee5be834

cleanup invalid prints (#18050)

calvin0327 committed 276 days ago

Verified 48545728

[Feature][V1] Support `tool_choice: required` when using Xgrammar as the `StructuredOutputBackend`. (#17845)

chaunceyjiang committed 276 days ago

Verified dc1a8217

[Bugfix] Avoid repeatedly creating dummy data during engine startup (#17935)

DarkLight1337 committed 276 days ago

Verified 61e0a506

[Bugfix] Fixes for new marlin moe usage (#18017)

mgoin committed 276 days ago

Verified 1df491c5

[ROCm]: Fix build from source failure with gcc14 and ROCm 6.3 (#13779)

arjunkathuria committed 277 days ago

Verified d8487ef5

[Misc] Slight spelling modification (#18039)

jeejeelee committed 277 days ago

Verified c06af9a9

Implements dual-chunk-flash-attn backend for dual chunk attention with sparse attention support (#11844)

sighingnow committed 277 days ago

Verified 60f76243

[ROCm] Skip tests for quantizations incompatible with ROCm (#17905)

hissu-hyvarinen committed 277 days ago

Verified f6518b2b

Remove noisy warnings from `SchedulerConfig` (#17995)

hmellor committed 277 days ago

Verified d67085c2

Use NVFP4 Marlin for CompressedTensorsW4A16Fp4 (#18000)

mgoin committed 277 days ago

Verified 307939f2

Update some more deprecated type hinting (#17998)

hmellor committed 277 days ago

Verified 9d7ea9db

[Model] Support MiMo-7B inference with MTP (#17433)

bwshen-mi committed 277 days ago

Verified acee8f48

Fix FBGEMM integration (#18002)

mgoin committed 277 days ago

Verified f065de4e

[V1][Spec Decode] Eagle unit tests (#17350)

wwl2755 committed 277 days ago

Verified dc990536

[CI] Make JSON output tests less likely to fail (#17859)

russellb committed 277 days ago

Verified ebab1ac3

Enable standard language model for torhc nightly (#18004)

yangw-dev committed 277 days ago

Verified 2b0db9b0

[Chore] Remove unused method (#18024)

robertgshaw2-redhat committed 277 days ago

Verified 195adb47

[v1][KVCacheManager] Change prefix caching metric from counting blocks to counting tokens (#18003)

heheda12345 committed 277 days ago

Verified 302f3aca

Enabling "Weight Loading Multiple GPU Test - Large Models" (#18020)

Alexei-V-Ivanov-AMD committed 277 days ago

Verified e9c730c9

[Core] Use platform-agnostic device control for DP engine core (#17245)

jianzs committed 277 days ago

Verified 289199fe

[CI/Build] Fix TPU V1 Test mixed use of & and && across tests (#17968)

CAROLZXYZXY committed 277 days ago

Verified b9fd0d7a

Construct `KVTransferConfig` properly from Python instead of using JSON blobs without CLI (#17994)

hmellor committed 277 days ago

Verified 72a3f6b8

[Lora][Frontend]Add default local directory LoRA resolver plugin. (#16855)

jberkhahn committed 277 days ago

Verified 98ea3560

[P/D] NIXL Integration (#17751)

robertgshaw2-redhat committed 277 days ago

Verified d1911020

Initialize the delta tool call fields explicitly (#17340)

maxdebayser committed 277 days ago

Verified 05a4324f

Newer Older