[CPU] Add fused GDN support for AMX CPU platform (#42707)

Li, Jiang committed 6 months ago

[Kernel] Pack topk id/weights triton kernel (#42527)

Jee Jee Li committed 6 months ago

[CPU] Specify required KV cache layout for CPU attention backend (#42740)

Tony Lin committed 6 months ago

delete xpu ci (#42582)

wenjun liu committed 6 months ago

[Perf] Add do_not_specialize in fused FP8 RoPE kernel (#42849)

Xin Yang committed 6 months ago

Revert checkpoint specific workaround in Transformers modelling backend (#42923)

Harry Mellor committed 6 months ago

[Model] [Perf] Use flatten for Qwen3.5's GDN output projection (#42311)

Rishapveer Singh committed 6 months ago

[LoRA] Support 2D and 3D MoE LoRA adapter at the same time (#42242)

Jee Jee Li committed 6 months ago

[Bugfix] moe lora align kernel grid (#40131)

Nguyễn Thế Duy committed 6 months ago

Improve logging when docs build is skipped (#42929)

Harry Mellor committed 6 months ago

[BugFix] Kimi-K2.5: skip vision tower dtype conversion when using quantization (#42869)

gaozihao-shy committed 6 months ago

[ROCm][CI] Stabilize ROCm pooling and multimodal CI (#42909)

Andreas Karatzas committed 6 months ago

[MM][CG] Enable encoder Cudagraph for Step3VL (#42224)

Soyaazz committed 6 months ago

[CI] Add NIXL EP import canary (#42567)

Alec committed 6 months ago

[Perf] Wire silu_and_mul_per_block_quant into TritonFP8MoE (MiniMax-M2) (#42497)

Yiliu Dong committed 6 months ago

[torch.compile] Add patch for fullgraph compilation (#42686)

Luka Govedič committed 6 months ago

[ROCm] [Bugfix] Fix DeepSeek V4 Functionality and Accuracy (#42810)

TJian committed 6 months ago

Refactor: Pass num_labels explicitly to PoolerClassify instead of reading from global config (#42851)

Taneem Ibrahim committed 6 months ago

[XPU] fix weight scale shape (#42725)

zofia committed 6 months ago

[MRV2][XPU] add Model Runner V2 log (#42710)

liuzhenwei committed 6 months ago