Commits

Branch selector

 main

User selector

vincentzed

All time

Commit History

Commits on Feb 12, 2026

• Avoid kimi linear stream sync (#16186)

  Show description for dc1309f
  vincentzed

  authored
  dc1309f

  Copy full SHA for dc1309f

Commits on Feb 8, 2026

• [ModelOPT] Support Qwen 3 Next Coder NVFP4 (#18224)

  Show description for bf89cc3
  vincentzed

  authored
  bf89cc3

  Copy full SHA for bf89cc3

• [ModelOpt] Fix broken Qwen3-235B-A22B-Instruct-2507-NVFP4 launch (#18189)

  Show description for ca36d88
  vincentzed

  authored
  ca36d88

  Copy full SHA for ca36d88

Commits on Feb 1, 2026

• [Performance] Optimize Mllama LayerNorm -> Upd (#9725)

  vincentzed

  authored
  c2ab371

  Copy full SHA for c2ab371

Commits on Jan 27, 2026

• Make flashMLA work on: Cu13, B300 (#17600)

  Show description for 8acd4d7
  vincentzed

  authored
  8acd4d7

  Copy full SHA for 8acd4d7

Commits on Jan 23, 2026

• add the fa4 mm backend and varlen func (#13539)

  Show description for 08fcda2
  

  vincentzed

  and

  b8zhong

  authored
  08fcda2

  Copy full SHA for 08fcda2

Commits on Jan 21, 2026

• [docs] Show user the fastAPI docs available (#17510)

  Show description for 458fe5a
  vincentzed

  authored
  458fe5a

  Copy full SHA for 458fe5a

Commits on Jan 16, 2026

• Add olmo3 in supported docs (#13672)

  Show description for ec9b48e
  vincentzed

  authored
  ec9b48e

  Copy full SHA for ec9b48e

Commits on Jan 15, 2026

• Show how to use cu13 image with B300 (#17170)

  Show description for 7dde343
  vincentzed

  authored
  7dde343

  Copy full SHA for 7dde343

• docs only add kimi k2 thinking and kimi linear (#15789)

  vincentzed

  authored
  d1110e1

  Copy full SHA for d1110e1

Commits on Jan 9, 2026

• Reduce some small cpu overhead in stream fetch (#16587)

  Show description for 84ab32a
  vincentzed

  authored
  84ab32a

  Copy full SHA for 84ab32a

Commits on Jan 8, 2026

• [perf] Add two stream norm for `Olmo3` speedup 5% (#13681)

  Show description for cda3561
  

  vincentzed

  and

  b8zhong

  authored
  cda3561

  Copy full SHA for cda3561

Commits on Dec 24, 2025

• Tiny env cleanup in deepgemm (#15706)

  Show description for ddb3970
  vincentzed

  authored
  ddb3970

  Copy full SHA for ddb3970

• Move some quant args to its own section in environ variables doc (#15722)

  Show description for ac320a6
  vincentzed

  authored
  ac320a6

  Copy full SHA for ac320a6

Commits on Dec 6, 2025

• fix: fix rmsnorm -> layernorm in qwen3 omni (#11791)

  Show description for 2ac5b98
  

  vincentzed

  and

  b8zhong

  authored
  2ac5b98

  Copy full SHA for 2ac5b98

Commits on Nov 20, 2025

• Fix launch of `Olmo3` (#13666)

  Show description for 6bc3062
  vincentzed

  authored
  6bc3062

  Copy full SHA for 6bc3062

Commits on Nov 15, 2025

• Clean up deprecated tile_tokens_dim for next flashinfer (#13341)

  Show description for 0d116b9
  vincentzed

  authored
  0d116b9

  Copy full SHA for 0d116b9

Commits on Oct 26, 2025

• [docs] upd docker files names everywhere (#12133)

  vincentzed

  authored
  b4d2da1

  Copy full SHA for b4d2da1

Commits on Oct 21, 2025

• [smol] [perf] Inverse perm improvement (#11482)

  Show description for e8640ee
  vincentzed

  authored
  e8640ee

  Copy full SHA for e8640ee

Commits on Oct 12, 2025

• [ perf ] Replace json-> orjson in hot path (#11221)

  Show description for a220536
  vincentzed

  authored
  a220536

  Copy full SHA for a220536

• [smol] [perf] Qwen3-VL in place op. (#11481)

  Show description for be740ac
  vincentzed

  authored
  be740ac

  Copy full SHA for be740ac

Commits on Oct 5, 2025

• Update `v1/responses` to be more OpenAI-compatible. (#9624)

  vincentzed

  authored
  36a6b8d

  Copy full SHA for 36a6b8d

Commits on Sep 29, 2025

• Fix gemma 3 launch with `transformers:` the error: `AttributeError: 'TransformersForCausalLM' object has no attribute 'tp_size'` (#9614)

  vincentzed

  authored
  7195954

  Copy full SHA for 7195954

Commits on Sep 15, 2025

• Add reasoning examples for GPT-OSS in Markdown examples (#9626)

  vincentzed

  authored
  0b14159

  Copy full SHA for 0b14159

• [docs / oneliner] update mmmu docs instruction (#9768)

  Show description for 1489cd6
  

  vincentzed

  and

  gemini-code-assist[bot]

  authored
  1489cd6

  Copy full SHA for 1489cd6

Commits on Aug 31, 2025

• [Performance] Improve Qwen RMSNorm by replacing with native RMSNorm op (#9709)

  vincentzed

  authored
  9a0d0b7

  Copy full SHA for 9a0d0b7

Commits on Aug 24, 2025

• fix(grok): remove duplicate replicate_lm_head configuration (#9549)

  vincentzed

  authored
  327f7b7

  Copy full SHA for 327f7b7

Commits on May 27, 2025

• docs: Update documentation to reflect xgrammar as default grammar backend (#6601)

  Show description for 45a31a8
  

  vincentzed

  and

  b8zhong

  authored
  45a31a8

  Copy full SHA for 45a31a8

Commits on May 17, 2025

• docs: Update the MD files (#6373)

  Show description for e9ef39d
  

  vincentzed

  and

  b8zhong

  authored
  e9ef39d

  Copy full SHA for e9ef39d

Commits on May 4, 2025

• Tool Call: Add `chat_template_kwargs` documentation (#5679)

  vincentzed

  authored
  95c231e

  Copy full SHA for 95c231e

Commits on Apr 27, 2025

• fix: Use `is not None` instead of `!= None` for None checks. (#5687)

  vincentzed

  authored
  094891c

  Copy full SHA for 094891c

• we fix the non existent access of `decrypted_config_file` (#5685)

  vincentzed

  authored
  df2cf58

  Copy full SHA for df2cf58