apunkt/MetaGPT
1
0
Fork 0
mirror of https://github.com/FoundationAgents/MetaGPT.git synced 2026-05-04 21:32:38 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions
1337 commits 28 branches 23 tags 341 MiB
Commit graph

129 commits

Author SHA1 Message Date
shenchucheng
b113aa246f update log_llm_stream in log_llm_stream.py/ollama_api.py 2023-12-25 17:22:30 +08:00
shenchucheng
b4552938e6 add llm stream log 2023-12-23 22:45:20 +08:00
better629
40d3cc5f81 format general_api_requestor params type 2023-12-22 09:51:26 +08:00
better629
4b0cb0084a add ollama support 2023-12-22 02:20:43 +08:00
geekan
4bf1844022
Merge pull request #596 from orange-crow/update-Message-instance 
Message(msg) -> Message(content=msg)
 2023-12-21 15:13:31 +08:00
geekan
a4843cd974
Merge pull request #595 from better629/feat_gemini 
Feat gemini
 2023-12-21 15:10:40 +08:00
better629
6af9fecf65 fix format 2023-12-21 15:06:59 +08:00
geekan
d46b7c4018 fix moderation, remove claude from LLM, refine exceptions handler 2023-12-21 14:45:53 +08:00
better629
9bd900452c fix conflicts 2023-12-21 14:29:01 +08:00
better629
bdb427d5b7 add gemini minimal python version warning 2023-12-21 14:18:50 +08:00
刘棒棒
44e648eabf Message(msg) -> Message(content=msg) 2023-12-21 14:17:05 +08:00
better629
f3eb9f638e add other llm for LLMProviderRegistry 2023-12-21 12:55:45 +08:00
seehi
163da9a2e7 format code 2023-12-21 12:44:43 +08:00
better629
e5a7fdfe3b retry use wait_random_exponential 2023-12-21 12:16:19 +08:00
better629
02090af7cb update gemini count_tokens 2023-12-21 12:16:19 +08:00
better629
91d1ab20cc update gemini user_msg doc 2023-12-21 12:16:19 +08:00
better629
c4fbc478d2 add google gemini 2023-12-21 12:16:17 +08:00
seehi
9a4f0d555c
Merge branch 'main' into feature-openai-v1 2023-12-21 12:06:12 +08:00
seehi
1564b1bf14 upgrade openai 1.3.5 to 1.6.0 2023-12-21 11:47:29 +08:00
voidking
4929e41f18 run pre-commit to find potential issues and fix them 2023-12-21 10:48:46 +08:00
geekan
f32f9c82e5 add llm provider registry 2023-12-20 14:04:45 +08:00
geekan
b43d8462de refine config 2023-12-20 14:02:47 +08:00
geekan
e8cb7991c4 openai_api: refine logic 2023-12-20 13:54:27 +08:00
better629
32af743b36 rm metagpt/utils/utils.py 2023-12-20 10:54:49 +08:00
better629
70cbfb1e48 retry use wait_random_exponential 2023-12-15 17:30:25 +08:00
better629
4127ef8570 update gemini count_tokens 2023-12-15 17:06:59 +08:00
geekan
ccecb45b13 resolve conflicts 2023-12-14 23:54:38 +08:00
莘权 马
290fb8b8d0 refactor: format 2023-12-14 21:10:46 +08:00
莘权 马
829dfd8997 feat: merge geekan:main 2023-12-14 20:34:04 +08:00
better629
9fb6e7c459 update gemini user_msg doc 2023-12-14 16:54:56 +08:00
better629
bef8d64193 add google gemini 2023-12-14 16:45:40 +08:00
better629
8c1b9db7d2
Merge branch 'main' into feat_repair_llmoutput 2023-12-12 00:27:18 +08:00
莘权 马
ec8c703c5a feat: merge geekan:main 2023-12-08 20:03:13 +08:00
莘权 马
9d84c8f047 feat: +SummarizeCode, refactor project_name 2023-12-08 10:34:57 +08:00
seehi
f4505d0e39 upgrade tiktoken to support azure 2023-12-06 16:23:43 +08:00
seehi
ad347e0717 upgrade tiktoken to support azure 2023-12-06 16:06:17 +08:00
seehi
a617aab65b azure client 2023-12-06 11:58:13 +08:00
seehi
0d8b9cdc89 support new openai package 2023-12-05 15:36:38 +08:00
seehi
09134c9c72 support new openai package 2023-12-05 15:27:57 +08:00
seehi
eaf531e0ac support new openai package 2023-12-05 11:26:54 +08:00
Sirui Hong
4006edac04
Merge pull request #485 from a-cid/exponential_backoff 
Fixed MetaGPT crashing due to openai.error.RateLimitError
 2023-12-04 10:59:32 +08:00
Sirui Hong
6feeac7d70
Merge pull request #444 from ShankHarinath/minor-edits 
Minor fixes
 2023-12-04 10:40:27 +08:00
Auster Cid
5ef3076f20 reimplemented retries following suggestions in OpenAI's cookbook 2023-11-29 12:04:05 -03:00
莘权 马
78548c2ddc feat: merge geekan:cli-etc 2023-11-28 18:16:50 +08:00
geekan
22288a342d 1. 动作优化 
1. SummarizeCode动作：用于基于代码进行总结，思考bug、逻辑、todo
  2. CodeReview动作优化：目前强制要求回答问题，有更高的成功率了
    1. 增加了LGTM/LBTM的回答，在LGTM时会及时停止，不重写代码
    2. 目前增加了设置中的参数code_review_k_times，与reflexion类似，设置为2
    3. 仍然有概率发生指令不遵循，尤其是会有比较高的概率发生同时review多个代码文件，还没想好怎么解决 #FIXME
  3. 增加了env到Action结构中，现在可以直接调用环境接口了
  4. WriteDesign：去除了对project_name的纠正代码，现在引导下可以一次生成对
    1. 修改了提示词中的##格式，改为了JSON格式
2. 数据结构
  1. Document的标准化：Env->Repo->Document，其中Document/Asset/Code都是Document
    1. 原用于检索的Document改为IndexableDocument
  2. Repo结构引入：用于Document装载与元数据装载
  3. RepoParser引入：写了一个简单的AST parser（后续可能要换tree-sitter），给出了整库symbol
  4. Env中增加了set/get/set_doc/get_doc接口，用于set/get单个变量或者一个Document。这个逻辑后续或许会进一步简化
3. 配置优化
  1. 默认更换为gpt-4-1106-preview，以获得最好的效果与成本
  2. 提供~/.metagpt作为配置最高优先级目录，从中读取config.yaml
  3. workspace可以灵活指定了，在config中配置
  4. project_name可以由命令行指定，并且改为由ProductManager生成
4. metagpt作为默认命令行，而非python startup.py
metagpt --help

metagpt --project-name game_2048 "make a 2048 game"
metagpt "make a 2048 game"

metagpt --project-name game_2048 --inc "将2048改为4096"

metagpt --project-name game_2048 --auto-inc "make a 2048 game"
  1. 使用新的METAGPT_ROOT生成方式，而非寻找git，以便cli安装
  2. 命令行由fire换为了typer，它会带来相对更好的体验
  3. project_name可以灵活指定了，在metagpt命令行输入中配置
5. 其他
  1. 现在支持多国语言了，中文已测试
  2. BossRequirement -> UserRequirement
  3. 大量错误文本的修正，增加了可读性
  4. 中量提示词优化，稍微提升了一些准确率
  5. 暂时屏蔽了LongtermMemory相关逻辑，这个逻辑底层调用了langchain的FAISS，会带来~5秒加载耗时
  6. 修复了安装包中的部分描述错误
  7. 去除了config中在openai_proxy设定时对base的重复修改，这个修改应该在openai初始化时发生
  8. 修复了JSON在中文存储时的特定问题，ensure_ascii=False
 2023-11-27 15:48:12 +08:00
better629
9a2ac792fe add __init__ 2023-11-23 01:55:56 +08:00
better629
642335317b add independent openllm and fireworks config fields, add llm output postprecess plugin 2023-11-23 01:46:14 +08:00
莘权 马
cda032948f refactor: pre-commit run --all-files 2023-11-22 16:26:48 +08:00
better629
322aef6f1d fix zhipuapi's key to init openai api_key when using openai sdk 2023-11-21 19:24:57 +08:00
garylin2099
f57355c889
Merge pull request #490 from orange-crow/feat_openai_ask_code_via_function 
Feat openai ask code via function.
 2023-11-21 17:59:35 +08:00