[9/n] Migrate attention and cache kernels to torch stable ABI (continued) (#43717)

Chris Leonard committed 5 months ago

[Refactor] Remove dead code (#43234)

Wentao Ye committed 5 months ago

[kv_offload] Skip decode-phase blocks in CPU offload (#43797)

Itay Etelis committed 5 months ago

[XPU] add scale transpose to prepare_fp8_moe_layer_for_xpu and bump up kernels (#43277)

Qiming Zhang committed 5 months ago

Adjust design around encoder_cudagraph_forward (#42288)

Weida Hong committed 5 months ago

[feat] add GlmgaProcessor specific logits in `glm4_1v.py` (#43575)

Jared Wen committed 5 months ago

[DSv4] Move mHC tilelang kernels & Don't use CustomOP in dsv4/nvidia (#43905)

Woosuk Kwon committed 5 months ago

[Misc][NUMA] Auto-bind to PCT priority cores on DGX B300 + widen EngineCore across shard NUMA nodes (#43270)

Vadim Gimpelson committed 5 months ago

[Rust Frontend] Add `/version` endpoint using engine-reported value (#43854)

Bugen Zhao committed 5 months ago

[Model]Support Step-3.7-Flash (#43859)

ltd0924 committed 5 months ago

[CI] Enable prefix caching in BFCL benchmark (#43925)

yzong-rh committed 5 months ago

fix(frontend): Add multimodal placeholders to Gemma4 tool message template (#41459)

Harshal Janjani committed 5 months ago

[AMD][CI][BugFix] Fix Distributed Compile Unit Tests (2xH100-2xMI300) group (#43120)

rasmith committed 5 months ago

Refactor output filename handling in ci-fetch-log.sh (#43901)

Michael Goin committed 5 months ago

[Spec Decode] Allow causal DFlash (#43445)

Benjamin Chislett committed 5 months ago

[Model Refactoring] Remove unncessary torch op registration for DSv4 (#43891)

Woosuk Kwon committed 5 months ago

[KV Offload] Add per-request offloading policy via `on_new_request` lifecycle hook (#43205)

Ronen Schaffer committed 5 months ago

[Core] Cleanup KVConnector handling with PP + fix MRV2 (#43732)

Nick Hill committed 5 months ago

[Feat] Add support for per GPU worker RDMA NIC selection (#42083)

Raj Joshi committed 5 months ago

Deprecate `JAISLMHeadModel` (#43784)

Harry Mellor committed 5 months ago