[10/n] Migrate cuda_view and silu_and_mul_per_block_quant kernels to torch stale ABI. (#44334)

Chris Leonard committed 5 months ago

[XPU][MoE] support block_fp8_moe on xpu (#42139)

zofia committed 5 months ago

[Rust Frontend] Skip loading multimodal processor if `--language-model-only` is specified (#44500)

Bugen Zhao committed 5 months ago

fix: keep DeepSeek V4 RoPE cache on inv_freq device (#43926)

Schwinn Saereesitthipitak committed 5 months ago

[mamba] unify KDA conv states into one cache to match 2-state SSM layout (#44539)

Jiangyun Zhu committed 5 months ago

[Logs Refactor] Optimize shutdown logs, easier to follow and consistent (#43707)

Wentao Ye committed 5 months ago

use split_group for pytorch process group creation (#41980)

Tushar Jain committed 5 months ago

[Kernel][Test] Extend lightning_attn and awq_triton kernel tests to XPU (#43307)

Agata Dobrzyniewicz committed 5 months ago

[Bugfix] Fix test_cutlass_moe.py (#44380)

bnellnm committed 5 months ago

[Misc] Add unit tests for pooler head classes (#44471)

Taneem Ibrahim committed 5 months ago

[DOC] Add INT8 W4A8 docs and Arm's supported quantization schemes (#34894)

Fadi Arafeh committed 5 months ago

[ROCm][CI] Add test for Aiter unified attn kernel (#44436)

Divakar Verma committed 5 months ago

[Bugfix] Reject non-positive values for ParallelConfig int knobs (#44057)

JianweiZheng committed 5 months ago

[Core] Freeze garbage collector in workers after model initialization (#44363)

Tyler Michael Smith committed 5 months ago

[Bugfix] MiniCPM-V-4.6 video inference crash: placeholder count mismatches visual embedding count (#44509)

tc-mb committed 5 months ago

Add model support for granite speech plus (#43519)

Zvi Kons committed 5 months ago

[Quant] Support compressed-tensors WNA8O8Int linears and WNInt embeddings (#44340)

Michael Goin committed 5 months ago

[DSv4] Adding TRTLLM gen attention kernel (#43827)

Yongye Zhu committed 5 months ago

[ROCm][CI] Specifying time outs for the lm eval models (#44255)

Andreas Karatzas committed 5 months ago

[ROCm][CI] Stabilize memory-release in the Hybrid model generation tests (#44046)

Andreas Karatzas committed 5 months ago