daisd-ai/cannary-re-reasoning-v2
The daisd-ai/cannary-re-reasoning-v2 is a 12 billion parameter language model fine-tuned from mistralai/Mistral-Nemo-Base-2407. It was trained on the biomedical_gpt_oss_full dataset, specializing it for biomedical reasoning and understanding. With a context length of 32768 tokens, this model is designed for applications requiring deep comprehension and generation within the biomedical domain.
Loading preview...
Model Overview
The daisd-ai/cannary-re-reasoning-v2 is a 12 billion parameter language model developed by daisd-ai. It is a specialized fine-tune of the mistralai/Mistral-Nemo-Base-2407 base model, leveraging its robust architecture for domain-specific applications. The model has been trained using the TRL library on the arynkiewicz/biomedical_gpt_oss_full dataset, indicating a strong focus on biomedical text understanding and generation.
Key Capabilities
- Biomedical Domain Specialization: Fine-tuned on a comprehensive biomedical dataset, making it suitable for tasks within life sciences and healthcare.
- Reasoning Tasks: Optimized for reasoning capabilities, likely within its specialized domain, given the base model's strengths and the fine-tuning dataset.
- Extended Context Window: Supports a context length of 32768 tokens, allowing for processing and understanding longer biomedical texts and complex queries.
Training Details
The model was trained using Supervised Fine-Tuning (SFT) with the TRL framework. The specific versions of libraries used include TRL 0.23.0, Transformers 4.56.1, PyTorch 2.6.0, Datasets 4.0.0, and Tokenizers 0.22.0.
Potential Use Cases
This model is well-suited for applications requiring advanced language understanding and generation in the biomedical field, such as:
- Biomedical question answering
- Scientific literature review and summarization
- Clinical text analysis
- Drug discovery research support