sparkle-reasoning/sparkle
[NeurIPS'25] Beyond Accuracy: Dissecting Mathematical Reasoning for LLMs Under Reinforcement Learning
This framework helps AI researchers and practitioners improve how large language models (LLMs) solve complex mathematical problems. By feeding LLMs specifically structured hard math problems with partial solutions, it teaches them to reason more effectively. The outcome is an LLM that can better understand, plan, and execute steps for solving math challenges, making it useful for those developing or deploying AI for quantitative tasks.
Use this if you are a machine learning researcher or engineer focused on advancing LLM capabilities for mathematical problem-solving through reinforcement learning.
Not ideal if you are a general user looking for a ready-to-use LLM for basic math or if you are not familiar with training and fine-tuning large language models.
Stars
16
Forks
—
Language
Python
License
MIT
Category
Last pushed
Dec 12, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/sparkle-reasoning/sparkle"
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