[Misc] Unify tokenizer registration (#29767)

Cyrus Leung committed 12 months ago

[crashfix] Eagle + multimodal can crash on mm cache miss (#29750)

Mickaël Seznec committed 12 months ago

[Frontend] Add tool filtering support to ToolServer (#29224)

daniel-salib committed 12 months ago

[Frontend] Resettle pooling entrypoints (#29634)

wang.yuqi committed 12 months ago

[CI] Skip paddleocr_vl for transformer 4.57.3 (#29758)

Huamin Li committed 12 months ago

Make PyTorch profiler gzip and CUDA time dump configurable (#29568)

Yifei Zhang committed 12 months ago

[MoE] CuteDSL MoE with Nvfp4 DeepEP dispatch (#27141)

Shu Wang committed 12 months ago

[Model Runner V2] Use packed mask for prompt bin counts (#29756)

Woosuk Kwon committed 12 months ago

[Misc]Remove redundant hidden_size property in ModelConfig (#29749)

Xingyu Liu committed 12 months ago

[Feat] Support non-gated activations in NVFP4 modelopt path (#29004)

Omer Ullman Argov committed 12 months ago

Fix RoPE failures in Transformers nightly (#29700)

Harry Mellor committed 12 months ago

[ROCm][Attention] Sliding window support for `AiterFlashAttentionBackend` (#29234)

Pleaplusone committed 12 months ago

[Core] Enable `inputs_embeds_size` separate from `hidden_size` (#29741)

Cyrus Leung committed 12 months ago

[Bugfix] Fix mismatched nvfp4 gemm output shape (#29742)

Isotr0py committed 12 months ago

[Misc] Update `TokenizerLike` interface and move `get_cached_tokenizer` (#29730)

Cyrus Leung committed 12 months ago

[Doc]: Fix typo in fused_moe layer (#29731)

朝 committed 12 months ago

[Bugfix][sleepmode][fp8 kv cache]: Fix FP8 KV cache + sleep(level=2) gibberish output (#28783)

Vensen committed 12 months ago

Fix AttributeError about _use_fi_prefill (#29734)

Huamin Li committed 12 months ago

[Quantization] Enable compressed-tensors AWQ for Turing GPU (#29732)

Isotr0py committed 12 months ago

[LoRA] Support FusedMoE LoRA Triton kernel for mxfp4 (#29708)

Xin Yang committed 12 months ago