xuyige/SoftCoT
ACL'2025: SoftCoT: Soft Chain-of-Thought for Efficient Reasoning with LLMs. and preprint: SoftCoT++: Test-Time Scaling with Soft Chain-of-Thought Reasoning
This project helps AI researchers and practitioners improve the reasoning capabilities and efficiency of large language models (LLMs). It takes an input prompt and an LLM, and outputs more accurate and robust answers by generating 'soft thoughts' that guide the LLM's reasoning process. This is for professionals who are developing or deploying LLM-powered applications and need to enhance their models' performance on complex tasks.
No commits in the last 6 months.
Use this if you are working with large language models and want to enhance their ability to reason through problems, especially in areas like mathematical problem-solving, without experiencing 'catastrophic forgetting' of prior knowledge.
Not ideal if you are looking for a simple, out-of-the-box solution for general LLM usage without delving into advanced model training and evaluation techniques.
Stars
81
Forks
14
Language
Python
License
—
Category
Last pushed
May 30, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/xuyige/SoftCoT"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
SimonAytes/SoT
Official code repository for Sketch-of-Thought (SoT)
Fr0zenCrane/UniCoT
[ICLR 2026] Uni-CoT: Towards Unified Chain-of-Thought Reasoning Across Text and Vision
logikon-ai/cot-eval
A framework for evaluating the effectiveness of chain-of-thought reasoning in language models.
nicolay-r/THOR-ECAC
The official fork of THoR Chain-of-Thought framework, enhanced and adapted for Emotion Cause...
andrewginns/CoT-at-Home
Who needs o1 anyways. Add CoT to any OpenAI compatible endpoint.