brucewayne0459/OpenBioLLm-Derm

TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:8kPublished:Jul 31, 2024License:mitArchitecture:Transformer0.0K Open Weights Cold

OpenBioLLm-Derm is an 8 billion parameter text generation model developed by Bruce_Wayne(The Batman), fine-tuned from aaditya/Llama3-OpenBioLLM-8B. This LLaMA-based model is specifically optimized for dermatology, providing clear, accurate, and helpful information about skin diseases, skincare routines, treatments, and related dermatological advice. It is designed for use in dermatology chatbots, offering specialized knowledge in this medical domain.

Loading preview...

OpenBioLLm-Derm: Specialized Dermatology LLM

OpenBioLLm-Derm is an 8 billion parameter language model developed by Bruce_Wayne(The Batman), fine-tuned from the LLaMA-based aaditya/Llama3-OpenBioLLM-8B. This model is specifically designed for applications in dermatology, leveraging its training on a dedicated dataset of skin diseases and care.

Key Capabilities

  • Dermatological Information: Provides clear, accurate, and helpful information on various skin diseases, skincare routines, treatments, and general dermatological advice.
  • Chatbot Integration: Primarily intended for use as the core intelligence for dermatology chatbots.
  • Specialized Knowledge: Offers in-depth knowledge within the dermatology domain, distinguishing it from general-purpose LLMs.

Training Details

The model was fine-tuned on a dataset focused on skin diseases and care (brucewayne0459/Skin_diseases_and_care). Training involved a per-device batch size of 2, 4 gradient accumulation steps, 120 max steps, and a learning rate of 2e-4 using the AdamW (8-bit) optimizer. The training was performed on a Tesla T4 GPU with 4-bit quantization and gradient checkpointing.

Limitations and Recommendations

Users should be aware that the model's responses are not a substitute for professional medical advice. It may have limitations in understanding rare conditions or those not well-represented in its training data. Further fine-tuning is recommended to enhance accuracy.