-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path2024-09-28.txt
25 lines (22 loc) · 2.93 KB
/
2024-09-28.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
2024-09-28
python
1. [meta-llama / llama-stack]:Llama Stack API 的模型组件(https://github.com/meta-llama/llama-stack)
2. [meta-llama / llama-stack-apps]:Llama Stack API 的代理组件(https://github.com/meta-llama/llama-stack-apps)
3. [meta-llama / llama-models]:用于 Llama 模型的实用工具(https://github.com/meta-llama/llama-models)
4. [hacksider / Deep-Live-Cam]:仅使用单张图片实现实时人脸交换和一键式视频深度伪造(https://github.com/hacksider/Deep-Live-Cam)
5. [microsoft / RD-Agent]:研究和开发(R&D)对于提高工业生产力至关重要,特别是在 AI 时代,R&D 的核心方面主要集中于数据和模型。我们致力于通过我们开源的 R&D 自动化工具 RD-Agent 自动化这些高价值通用 R&D 流程,让 AI 驱动数据驱动 AI。(https://github.com/microsoft/RD-Agent)
6. [stanfordnlp / dspy]:DSPy:用于编程(而非提示)基础模型的框架(https://github.com/stanfordnlp/dspy)
7. [explodinggradients / ragas]:用于评估您的检索增强生成(RAG)管道的评估框架(https://github.com/explodinggradients/ragas)
8. [unclecode / crawl4ai]:🔥🕷️ Crawl4AI:开源的 LLM 友好型网络爬虫和抓取工具(https://github.com/unclecode/crawl4ai)
9. [huggingface / trl]:使用强化学习训练 Transformer 语言模型(https://github.com/huggingface/trl)
10. [hyperknot / openfreemap]:免费开源的地图托管解决方案,为网站和应用提供自定义样式,使用 OpenStreetMap 数据(https://github.com/hyperknot/openfreemap)
11. [odoo / odoo]:Odoo。开源应用,助力您的企业成长。(https://github.com/odoo/odoo)
12. [simonw / llm]:从命令行访问大型语言模型(https://github.com/simonw/llm)
13. [Sinaptik-AI / pandas-ai]:与您的数据库(SQL、CSV、pandas、polars、mongodb、noSQL 等)聊天。PandasAI 使用 LLM(GPT 3.5 / 4、Anthropic、VertexAI)和 RAG 使数据分析对话化。(https://github.com/Sinaptik-AI/pandas-ai)
14. [openai / tiktoken]:tiktoken 是用于与 OpenAI 模型一起使用的快速 BPE 分词器(https://github.com/openai/tiktoken)
15. [cpacker / MemGPT]:Letta(原名 MemGPT)是一个用于创建状态化 LLM 服务的框架。(https://github.com/cpacker/MemGPT)
16. [TheAlgorithms / Python]:所有算法在 Python 中实现(https://github.com/TheAlgorithms/Python)
17. [meta-llama / llama]:Llama 模型的推理代码(https://github.com/meta-llama/llama)
18. [meta-llama / llama3]:Meta Llama 3 的官方 GitHub 网站(https://github.com/meta-llama/llama3)
最惊艳的项目推荐:[hacksider / Deep-Live-Cam] - 仅使用单张图片实现实时人脸交换和一键式视频深度伪造。
今日趋势项目关注的领域和特点:主要集中在人工智能和机器学习领域,尤其是深度学习和自然语言处理。项目特点包括创新性技术、开源和跨平台应用,以及对于现有技术的优化和扩展。