Qiaolin Yu 09a118fafe Support return_logprob for spec v2 (overlap safe) (#19801) ...

Co-authored-by: Ratish1 <ratish1501@gmail.com>
Co-authored-by: Ratish1 <formula733@gmail.com>
Co-authored-by: hnyls2002 <lsyincs@gmail.com>

2026-03-10 15:38:27 -07:00

..

4-gpu-models

[Qwen3.5] Enable MTP spec_v2 and add test for nvidia/Qwen3.5-397B-A17B-NVFP4 (#19391)

2026-03-04 14:01:25 -08:00

8-gpu-models

[Nightly] Replace MiniMax-M2 with MiniMax-M2.5 (#20083)

2026-03-07 01:15:34 -08:00

amd

[AMD] [Z-Image-Turbo Day 0] Add Z-Image-Turbo nightly test for AMD GPUs (#19733)

2026-03-05 08:36:17 -08:00

ascend

[NPU]grok2 model support (#17119)

2026-03-03 10:24:10 +08:00

attention

fix flaky test for test_kda_kernels (#19864)

2026-03-04 22:47:29 +08:00

backends

[FlashInfer v0.6.4] [RL] Integrate FlashInfer mxfp8 gemm, MoE, and routed MoE (#19537)

2026-03-10 15:37:57 -07:00

bench_fn

[Benchmark] Fix generated_shared_prefix attribute naming and remove args dependency (#19363)

2026-02-25 18:45:54 -08:00

constrained_decoding

[CI] Reorganize stage-b 1-GPU tests for 5090 compatibility (#16826)

2026-01-15 15:23:35 -08:00

core

[CI] Add GPT-OSS test for SM120 (#20056)

2026-03-06 11:43:04 -08:00

debug_utils

Enhance error resilience in dump comparator (#19685)

2026-03-02 19:08:35 +08:00

disaggregation

[Performance] Decode Offload improves the long texts performance 100% through dynamic block offload. (#17216)

2026-03-08 17:16:53 +08:00

distributed

[Qwen3.5] Support Qwen3.5 Pipeline Parallelism (#19670)

2026-03-07 23:34:08 +08:00

dllm

[DLLM] Add initial radix cache support (#18724)

2026-03-04 23:24:09 -08:00

ep

chore: bump mooncake version to 0.3.9 (#18316)

2026-02-07 17:30:01 +08:00

eval

Fix nightly VLM accuracy: gemma3n TP fixes + removal, latency thresholds (#19401)

2026-02-27 14:24:02 -08:00

function_call

[Tool Call] Stream DeepSeek-V3.2 function call parameters in JSON format. (#16091)

2026-03-03 01:46:29 -08:00

hicache

[Benchmark] Remove re-exports from bench_serving.py (#19130)

2026-02-21 14:30:30 -08:00

kernels

[DSv32] [GLM5] Improve Model Quality by Avoiding FP32 Precision Loss in weights_proj (#19041)

2026-02-22 16:20:51 +08:00

layers

Return intermediate Mamba states (#19716)

2026-03-09 16:04:36 +08:00

lora

[Bug] Fix lora tp bug on H200 (#19769)

2026-03-04 20:11:02 -08:00

metrics

[Bug] Fix missing TTFT histogram for single-batch requests (#20122)

2026-03-08 15:18:51 -07:00

mla

Relax flaky test thresholds for MLA DeepSeek V3 and AutoRound (#20068)

2026-03-06 17:26:28 -08:00

model_loading

Fix/nemotron mtp quantaized (#19433)

2026-03-03 01:07:46 -08:00

models

[AMD] Fix AMD CI - stage-b-small-1-gpu-amd (partition 7) (#20028)

2026-03-06 23:49:14 -08:00

moe

Merge stage-c-test-large-4-gpu suites into partitioned suites (#18325)

2026-02-06 15:32:33 -08:00

openai_server

fix: Properly return abort error for streaming requests if the abort is triggered by scheduler (#19357)

2026-03-03 17:18:15 -08:00

ops

[AMD] CI - Add MI35x nightly/PR tests for kv-cache-fp8 and allreduce-fusion (DeepSeek) (#19834)

2026-03-05 07:09:57 -08:00

parser

Add KimiK2Detector with tool interruption support (#19696)

2026-03-03 14:04:49 +08:00

perf

[TestFix] change LoRA tests to use NVIDIA adapter instead of Nutanix (#19642)

2026-03-02 12:55:41 -08:00

piecewise_cuda_graph

[CI] Refactor PCG related CI (#19994)

2026-03-10 14:49:25 +08:00

profiling

Migrate 4-GPU/8-GPU workflow jobs to stage-c and add CI registry decorators (#17299)

2026-01-31 22:37:22 -08:00

quant

[FlashInfer v0.6.4] [RL] Integrate FlashInfer mxfp8 gemm, MoE, and routed MoE (#19537)

2026-03-10 15:37:57 -07:00

radix_cache

[Feature] Add SLRU eviction policy & fix RadixCache hit_count bug (#18843)

2026-03-08 21:30:55 +08:00

rl

[feat] Enhance lora_update_weight_from_tensor for RL training (#19314)

2026-03-04 18:10:42 -08:00

rotary

feat: Support mrope_section with rope_type: "yarn" (#13313)

2026-02-14 22:51:44 +08:00

sampling

Fix flaky penalty tests by using higher temperature for effect comparison (#18380)

2026-02-12 21:08:37 +08:00

scheduler

add mixed chunk unit test and make small refactors (#18776)

2026-03-08 19:56:09 +08:00

sessions

Enable piecewise-cuda-graph when logprob_start_len = -1 (#19453)

2026-03-10 12:50:57 -07:00

spec

Support return_logprob for spec v2 (overlap safe) (#19801)

2026-03-10 15:38:27 -07:00

stress

Add stress test workflow (#13937)

2025-11-26 13:24:39 -08:00

tokenizer

[CI] Reorganize stage-b 1-GPU tests for 5090 compatibility (#16826)

2026-01-15 15:23:35 -08:00

utils

Decouple scheduler log printing from metrics collection (#20107)

2026-03-07 22:09:10 -08:00

vlm

[VLM] Replace decord with torchcodec for video decoding (#20055)

2026-03-09 19:23:49 +08:00

test_hybrid_dp_ep_tp_mtp.py

Add hybrid parallelism test to nightly CI (#17444)

2026-01-20 17:43:50 -08:00

test_srt_backend.py

[AMD CI] Migrate and Add More Testcases (#17116)

2026-01-19 08:07:39 -08:00