fix(security): bound tokenizer work when explicit truncation_side is set (#47007)

Juan Pérez de Algaba committed 4 months ago

[Model Runner V2][Spec Decode] Implement block verification for rejection sampling (#46781)

Giancarlo Delfin committed 4 months ago

[Bugfix][Quant] Raise actionable error instead of bare assert for group-size/TP mismatch (#46230) (#46236)

Arsalan Shakil committed 4 months ago

[Bugfix] Seed RayExecutorV2 TCPStore port by DP rank to avoid collisions (#45960)

Seiji Eicher committed 4 months ago

[Rust Frontend] Simplify unit tests with shared `TestTokenizer` (#47125)

Bugen Zhao committed 4 months ago

Remove more unnecessary `load_weights` methods (#47058)

Harry Mellor committed 4 months ago

[KV Offload] Pass `ScheduleEndContext` to `on_schedule_end` hook (#46450)

Ronen Schaffer committed 4 months ago

Forward fix nightly errors from #44589 (#47151)

Harry Mellor committed 4 months ago

[CI/Build] Bump PyNvVideoCodec version (#47139)

Isotr0py committed 4 months ago

[XPU] C++ implementation for get_memory_info (#47134)

Qiming Zhang committed 4 months ago

[Rust Frontend] Refactor TLS serve path with unified `MaybeTlsListener` (#47101)

Bugen Zhao committed 4 months ago

[Rust Frontend] Extend renderer/parser roundtrip tests to support token ids (#47110)

Bugen Zhao committed 4 months ago

[Model] Remove Tarsier, Tarsier2 (#47143)

Harry Mellor committed 4 months ago

[Bugfix][ROCm][MLA] Pass q/kv dtypes to get_mla_metadata_v1 in FP8 decode (#46997)

peizhang56 committed 4 months ago

Bump actions/checkout from 6.0.1 to 7.0.0 (#33057)

dependabot[bot] committed 4 months ago

[Bugfix] Reject negative values for max_logprobs and long_prefill_token_threshold (#44002)

JianweiZheng committed 4 months ago

[XPU][CI] Enable shared loader test (#45977)

Chaojun Zhang committed 4 months ago

[Attention Backend] add HPC-Ops Attention backend (#46020)

Cheng Jiang committed 4 months ago

[Kernel][XPU] Adjust kernel unit tests for XPU (#45140)

Agata Dobrzyniewicz committed 4 months ago

fix(security): prevent image decompression bomb OOM denial of service (#47010)

Juan Pérez de Algaba committed 4 months ago