l3lab

/

L1-Qwen-1.5B-Exact

like 6

Follow

L3 Lab 22

This PR significantly enhances the model card for the L1 model by:

• Adding pipeline_tag: text-generation to enable discovery and categorize the model correctly.
• Including library_name: transformers metadata, as evidenced by config.json and tokenizer_config.json (Qwen2ForCausalLM, LlamaTokenizer), which will enable the automated "how to use" widget on the model page.
• Providing a concise description of the model based on its abstract.
• Linking to the paper: L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning.
• Adding links to the project page (https://cmu-l3.github.io/l1) and the GitHub repository (https://github.com/cmu-l3/l1).
• Incorporating a Python code snippet for sample usage with the transformers library, extracted from the official Colab notebook, demonstrating text generation with length control using chat templates.
• Including the Acknowledgments and Citation sections from the original GitHub README for proper attribution.