Commits

Branch selector

 main

User selector

All users

All time

Commit History

Commits on Nov 21, 2025

• Enhance Windows CUDA path detection logic

  Show description for 4a30503
  JamePeng

  committed
  4a30503

  Copy full SHA for 4a30503

• Sync cmake : add option to build and link BoringSSL and OpenSSL

  JamePeng

  committed
  01ed775

  Copy full SHA for 01ed775

• Update Submodule vendor/llama.cpp 054a45c..23bc779

  JamePeng

  committed
  67421d5

  Copy full SHA for 67421d5

Commits on Nov 20, 2025

• Bump version to 0.3.17

  JamePeng

  committed
  8771222

  Copy full SHA for 8771222

• Update Submodule vendor/llama.cpp 4c91f26..054a45c

  JamePeng

  committed
  fbb0ed2

  Copy full SHA for fbb0ed2

• Update README.md for Qwen3VL example

  Show description for 681f171
  JamePeng

  committed
  681f171

  Copy full SHA for 681f171

• Add Pillow process code in _load_image which can reliably handle common formats

  Show description for 3b01333
  JamePeng

  committed
  3b01333

  Copy full SHA for 3b01333

• Optimize: Initialize Jinja template in __init__ instead of __call__ to avoid redundant parsing

  Show description for 30ace25
  JamePeng

  committed
  30ace25

  Copy full SHA for 30ace25

• Update pyproject.toml dependencies and requires-python>=3.9

  JamePeng

  committed
  baa60ae

  Copy full SHA for baa60ae

• Update Submodule vendor/llama.cpp 07b0e7a..4c91f26

  JamePeng

  committed
  35a100d

  Copy full SHA for 35a100d

Commits on Nov 19, 2025

• Update Submodule vendor/llama.cpp cb623de..07b0e7a

  JamePeng

  committed
  917cc0b

  Copy full SHA for 917cc0b

Commits on Nov 18, 2025

• Update Llava15ChatHandler to accept `use_gpu`, `image_min_tokens`, and `image_max_tokens`.

  Show description for 47d098e
  JamePeng

  committed
  47d098e

  Copy full SHA for 47d098e

Commits on Nov 17, 2025

• Update README.md

  JamePeng

  committed
  a32aa9e

  Copy full SHA for a32aa9e

• Update Submodule vendor/llama.cpp c7b7db0..cb623de

  JamePeng

  committed
  c51b589

  Copy full SHA for c51b589

• Add a Basic workflow for cu128 windows wheels

  JamePeng

  committed
  cfb4f8f

  Copy full SHA for cfb4f8f

Commits on Nov 16, 2025

• Supplementing the use of mtmd_helper_log_set to align with llama.cpp

  JamePeng

  committed
  e0dbf82

  Copy full SHA for e0dbf82

• Remove parameters that are no longer needed: mctx_params.verbosity

  JamePeng

  committed
  65eced9

  Copy full SHA for 65eced9

Commits on Nov 15, 2025

• Update llama.cpp 20251115 and Move the ggml-related code to _ggml.py.

  JamePeng

  committed
  77b1efb

  Copy full SHA for 77b1efb

• Update Submodule vendor/llama.cpp 017ecee..c7b7db0

  JamePeng

  committed
  6b5f4c6

  Copy full SHA for 6b5f4c6

Commits on Nov 12, 2025

• Use httplib to download model from an URL when the libcurl is disabled

  Show description for 4e901b7
  JamePeng

  committed
  4e901b7

  Copy full SHA for 4e901b7

• Update Submodule vendor/llama.cpp 7d019cf..017ecee

  JamePeng

  committed
  8cf35b2

  Copy full SHA for 8cf35b2

Commits on Nov 11, 2025

• Fixed CMake Error at vendor/llama.cpp/tools/mtmd/CMakeLists.txt:16 (set_target_properties)

  Show description for 326a4f3
  JamePeng

  committed
  326a4f3

  Copy full SHA for 326a4f3

• Update Submodule vendor/llama.cpp 0c74f32..7d019cf

  JamePeng

  committed
  62bc43b

  Copy full SHA for 62bc43b

Commits on Nov 10, 2025

• Refining the AVX instruction compilation workflow

  JamePeng

  committed
  dd38ec6

  Copy full SHA for dd38ec6

• Update LlamaContext API and Release the model pointer when the ctx was failed to create context with model

  JamePeng

  committed
  528a518

  Copy full SHA for 528a518

• Update Submodule vendor/llama.cpp 299f5d7..0c74f32

  JamePeng

  committed
  8b62fb0

  Copy full SHA for 8b62fb0

Commits on Nov 9, 2025

• Adding workflow compilation options for platforms without AVX instructions or AVXVNNI instruction can be achieved by setting the releasetag parameter above to Basic/AVXVNNI.

  Show description for 5abcf51
  JamePeng

  committed
  5abcf51

  Copy full SHA for 5abcf51

Commits on Nov 8, 2025

• Merge pull request #18 from TAO71-AI/main

  Show description for 3d96053
  JamePeng

  authored
  3d96053

  Copy full SHA for 3d96053

• Merge branch 'JamePeng:main' into main

  alcoftTAO

  authored
  04656b1

  Copy full SHA for 04656b1

• Fixed a typo in the Qwen3-VL chat template.

  alcoftTAO

  committed
  ade9fc8

  Copy full SHA for ade9fc8

• Update README.md

  JamePeng

  committed
  67baa01

  Copy full SHA for 67baa01

• Merge pull request #17 from TAO71-AI/main

  Show description for 17ba24f
  JamePeng

  authored
  17ba24f

  Copy full SHA for 17ba24f

• Added 'add_vision_id' to the chat template.

  alcoftTAO

  committed
  14d14cc

  Copy full SHA for 14d14cc

• Deleted 'thinking_budget' because it's not implemented yet.

  alcoftTAO

  committed
  a749dfa

  Copy full SHA for a749dfa

• Merge branch 'JamePeng:main' into main

  alcoftTAO

  authored
  48d6507

  Copy full SHA for 48d6507