fix(kimi_k25): resolve media_placeholder_token_id from tokenizer (#39344)

r266-tech committed 7 months ago

Support FP8 KVCache on XPU (#37731)

Xinyu Chen committed 7 months ago

[ROCm][CI/Build] Fix memory cleanup in MM test (#39555)

Andreas Karatzas committed 7 months ago

[XPU][CT] support per-channel quantization in xpu fp8 linear method (#38316)

Yan Ma committed 7 months ago

[Quant] add CompressedTensorsW8A8Mxfp8 for linear and MoE layers (#38815)

EdalatiAli committed 7 months ago

[Bugfix] Runtime driver check for cuMemcpyBatchAsync in swap_blocks_batch (#38919)

Itay Etelis committed 7 months ago

[Perf] Fuse Zero Initializer for FP8 DeepGemm Block Quant Kernel (#39547)

Wei Zhao committed 7 months ago

[Gemma4][Bugfix]: Enable Gemma4ForCasualLM to load lora adapters correctly (#38844)

ShubyM committed 7 months ago

[Bugfix] Fix GDN FLA kernel crashes with NULL_BLOCK_ID=0 CUDA graph padding (#39064)

Vibhav Agarwal committed 7 months ago

[CI/Build] Fix sentence-transformers version in CPU test (#39557)

Li, Jiang committed 7 months ago

[Misc] Update deprecation warning for --model flag (#39518)

z1ying committed 7 months ago

[Bugfix] add SupportsMultiModal to Exaone4_5_MTP (#39526)

Lee Yongjun committed 7 months ago

Fix the order of _free_encoder_inputs (#38907)

Tianyu Guo committed 7 months ago

[XPU] Fix spec-decode UTs under tests/v1/spec_decode (#38491)

Yan Ma committed 7 months ago

Revert "Add nightly b200 test for spec decode eagle correctness (#38577)" (#39512)

Benjamin Chislett committed 7 months ago

[Docs] Use `--torch-backend=auto` for editable install docs (#39511)

Michael Goin committed 7 months ago

[Refactor] Move MXFP8 GEMM management into MxFp8LinearKernel (#39205)

Michael Goin committed 7 months ago

Add structure to `requirements/` directory (#39024)

Harry Mellor committed 7 months ago

Fix pre-commit labeled trigger system (#39523)

Fynn Schmitt-Ulms committed 7 months ago

Add Gemma4 Eagle3 support (#39450)

Fynn Schmitt-Ulms committed 7 months ago