Demo for "Self-Training Elicits Concise Reasoning in Large Language Models"

This Space showcases the model tergel/llama-3.2-3b-instruct-gsm8k-fs-gpt4o-bon

We provide a simple chat interface allowing you to observe the concise CoT solutions that our model can produce. Feel free to play with it.

1 2048
0.1 4
0.05 1
1 1000
1 2