Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 15, 2024

• 更改Jinja elif逻辑，修复Qwen2.5多轮；支持Jinja负数下标，支持Qwen1.5最初版本

  

  cgli

  committed
  603911b

• 修复Qwen一代推理错误

  

  cgli

  committed
  ee2b970

Commits on Dec 13, 2024

• 修改一下调度

  

  黄宇扬

  committed
  2ed9993

• 增加支持TeleAI的telechat2-3B, 7B

  

  黄宇扬

  committed
  24d73d8

Commits on Dec 11, 2024

• 支持gemma2

  

  黄宇扬

  committed
  60871ce

Commits on Dec 9, 2024

• Merge branch 'master' of https://www.github.com/ztxz16/fastllm

  

  黄宇扬

  committed
  a51dea3

• 完成minicpm3的graph形式

  

  黄宇扬

  committed
  5e27fb7

• Merge pull request ztxz16#497 from TylunasLi/doc

  Show description for d22dd1d
  ztxz16

  authored
  d22dd1d

Commits on Dec 5, 2024

• Jinja2模板支持取模运算和数组切片，支持直接加载codellama-7b-instruct的HF模型

  

  cgli

  committed
  1650218

• 修复qwen批量推理错误

  

  cgli

  committed
  3b2220d

Commits on Dec 2, 2024

• 加速graph llm，达到和llama基本相同的速度

  

  黄宇扬

  committed
  f711b32

Commits on Dec 1, 2024

• 支持直接加载Qwen（一代）的HF模型

  

  cgli

  committed
  8c7eb27

• 支持MiniCPM3-4B导出并使用C++ Tokenizer

  

  cgli

  committed
  e6d4e56

Commits on Nov 23, 2024

• Jinja模板支持Not和括号条件，支持Qwen2.5

  

  cgli

  committed
  f9ea416

Commits on Oct 24, 2024

• cogvlm支持float16

  

  黄宇扬

  committed
  d295968

• add cudacat

  

  黄宇扬

  committed
  e6a5833

Commits on Oct 23, 2024

• Merge branch 'master' of https://www.github.com/ztxz16/fastllm

  

  黄宇扬

  committed
  62b3d05

• 初步支持cogvlm2

  

  黄宇扬

  committed
  f43c984

• 加一个简单的cpu conv2d

  

  黄宇扬

  committed
  c0708dc

Commits on Oct 15, 2024

• 修复不同device之间使用CopyFrom会出错的bug

  huangsheng-tf

  committed
  f6c25e0

• MiniCPM3更新

  huangsheng-tf

  committed
  26ff307

Commits on Oct 11, 2024

• 支持MiniCPM3模型

  huangsheng-tf

  committed
  31c4953

Commits on Sep 26, 2024

• 修复graphllm的内存泄漏

  

  黄宇扬

  committed
  a6c6c43

• 支持f16, bf16的lora

  

  黄宇扬

  committed
  7502e4c

• fix xlmberta

  

  黄宇扬

  committed
  3d61022

Commits on Sep 25, 2024

• 暂时注释xlmroberta中的attention加速

  

  黄宇扬

  committed
  5912fd3

• expandDims时清零，修复一些情况下出现！！！！的bug

  

  黄宇扬

  committed
  e0e9a6b

Commits on Sep 24, 2024

• xlmroberta speedup

  

  黄宇扬

  committed
  e883fc5

Commits on Sep 23, 2024

• add gelu for cuda

  

  黄宇扬

  committed
  02de743

• BERT类修改为一个支持embedding, reranker的基类

  

  黄宇扬

  committed
  016eb32

• fix 直接读bge-largh-zh

  

  黄宇扬

  committed
  4736a6d

Commits on Sep 22, 2024

• 修改embedding模型和reranker模型的python接口

  

  黄宇扬

  committed
  5b89fab

• fix

  

  黄宇扬

  committed
  0222fa1

• 支持xlmroberta模型

  

  黄宇扬

  committed
  72e5cba

Commits on Sep 20, 2024

• 增加embed python api接口，bert模型增加normalize

  jiewlmrh

  committed
  04aa027

Pagination

PreviousNext