AI Disruption

AI Disruption

Share this post

AI Disruption
AI Disruption
30 Latest AI Open Source Projects of the Week(2025.1.20-2025.1.26)
Copy link
Facebook
Email
Notes
More

30 Latest AI Open Source Projects of the Week(2025.1.20-2025.1.26)

Explore the latest AI open-source models and frameworks (Jan 20-26, 2025), including cutting-edge projects like Art v0, ScribeAgent, XiYan-SQL, and more.

Meng Li's avatar
Meng Li
Jan 27, 2025
∙ Paid
1

Share this post

AI Disruption
AI Disruption
30 Latest AI Open Source Projects of the Week(2025.1.20-2025.1.26)
Copy link
Facebook
Email
Notes
More
1
Share

"AI Disruption" publication New Year 30% discount link.


I’m sharing some interesting AI open-source models and frameworks from this week (2025.1.20-2025.1.26).

A total of 30 AI open-source projects.

Project: Art v0

Auto-Regressive Thinker (Art) v0 3B is the first model in the Art series, fine-tuned based on Qwen/Qwen2.5-3B-Instruct and trained using a dedicated dataset generated by Gemini 2.0 Flash Thinking.

The model uses a Transformer architecture, has 3B parameters, and focuses on enhancing reasoning abilities.

https://huggingface.co/AGI-0/Art-v0-3B

Project: ScribeAgent

data processing pipeline

The ScribeAgent project aims to fine-tune large language models (LLM) into specialized web agents by using production-scale workflow data.

This project fine-tunes using the publicly available Scribe dataset to improve the specialization capabilities of web agents.

The project provides code for data preprocessing, fine-tuning open-source LLMs, and benchmark evaluations with Mind2Web and WebArena.

https://github.com/colonylabs/ScribeAgent

Project: XiYan-SQL

image

XiYan-SQL is an innovative framework designed to enhance the candidate generation ability for natural language to SQL tasks using a multi-generator integration strategy.

This project introduces M-Schema, a semi-structured schema representation method, to improve the understanding of database structures. By combining contextual learning and supervised fine-tuning, XiYan-SQL improves the quality and diversity of generated SQL queries.

Experimental results show that this framework performs exceptionally well on multiple dialect datasets, significantly improving SQL query execution accuracy.

https://github.com/XGenerationLab/XiYan-SQL

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2025 Meng Li
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture

Share

Copy link
Facebook
Email
Notes
More