StringNLPLAB/MGS
Repository for the paper "Advancing General-Purpose Reasoning Models with Modular Gradient Surgery"
This project helps AI researchers and practitioners improve Large Language Models (LLMs) by balancing multiple training objectives like mathematical reasoning, general conversation, and instruction following. It takes an existing LLM and training data from diverse sources (e.g., math problems, chat logs), and outputs a more versatile LLM capable of strong performance across these different domains. This tool is for those who are fine-tuning or training LLMs for multi-skill applications.
Use this if you need to train a single LLM that performs well across distinct tasks such as complex mathematical reasoning, general chat, and accurately following instructions, without sacrificing performance in any one area.
Not ideal if you are looking for a pre-trained, ready-to-use LLM for a single, highly specialized task, or if you don't have the technical expertise to fine-tune advanced models.
Stars
19
Forks
—
Language
Python
License
MIT
Category
Last pushed
Mar 15, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/StringNLPLAB/MGS"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
cvs-health/uqlm
UQLM: Uncertainty Quantification for Language Models, is a Python package for UQ-based LLM...
PRIME-RL/TTRL
[NeurIPS 2025] TTRL: Test-Time Reinforcement Learning
sapientinc/HRM
Hierarchical Reasoning Model Official Release
tigerchen52/query_level_uncertainty
query-level uncertainty in LLMs
reasoning-survey/Awesome-Reasoning-Foundation-Models
✨✨Latest Papers and Benchmarks in Reasoning with Foundation Models