mlfoundations-dev/qwen_openthoughts_science_claude

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kPublished:Mar 25, 2025License:apache-2.0Architecture:Transformer Open Weights Warm

The mlfoundations-dev/qwen_openthoughts_science_claude model is a 7.6 billion parameter language model fine-tuned from Qwen/Qwen2.5-7B-Instruct. It was trained on the mlfoundations-dev/open-thoughts-science-claude dataset, suggesting a specialization in scientific reasoning or discourse. With a context length of 32768 tokens, it is designed for processing extensive scientific texts and complex problem-solving, making it suitable for research assistance and knowledge extraction in scientific domains.

Loading preview...

Overview

This model, mlfoundations-dev/qwen_openthoughts_science_claude, is a 7.6 billion parameter language model derived from the Qwen2.5-7B-Instruct architecture. It has been specifically fine-tuned using the mlfoundations-dev/open-thoughts-science-claude dataset.

Key Capabilities

  • Specialized Fine-tuning: Optimized through fine-tuning on a science-focused dataset, indicating potential strengths in scientific understanding and generation.
  • Base Model Heritage: Benefits from the robust capabilities of the Qwen2.5-7B-Instruct base model.
  • Extensive Context Window: Supports a context length of 32768 tokens, enabling the processing and analysis of long scientific documents and complex information.

Good For

  • Scientific Research: Assisting with tasks requiring deep understanding or generation of scientific content.
  • Knowledge Extraction: Extracting information from lengthy scientific papers or technical reports.
  • Complex Problem Solving: Engaging with intricate scientific problems that benefit from a large context window.

Training Details

The model was trained with a learning rate of 8e-05, a total batch size of 512 (achieved with gradient accumulation steps of 32), and for 3 epochs. The optimizer used was adamw_torch with a cosine learning rate scheduler and a warmup ratio of 0.1.