AmberYifan/capsd-medicine-dedup-marin-8b-base-medicine_cap_b2000_s0

Model Overview

This model, AmberYifan/capsd-medicine-dedup-marin-8b-base-medicine_cap_b2000_s0, is an 8 billion parameter language model derived from the marin-community/marin-8b-base architecture. It has been specifically fine-tuned on a dedicated medical dataset, capsd_marin-8b-base-n80000-medicine-dedup80k__mix_medicine_cap_b2000_s0, to specialize its capabilities within the healthcare domain. The training process involved a single epoch with a learning rate of 1e-05 and a total batch size of 64, utilizing a cosine learning rate scheduler.

Key Characteristics

• Base Model: Fine-tuned from marin-community/marin-8b-base.
• Parameter Count: 8 billion parameters.
• Context Length: Supports an 8192-token context window.
• Specialized Training: Underwent fine-tuning on a deduplicated medical dataset, indicating an optimization for medical-related language tasks.

Training Details

• Optimizer: AdamW with betas=(0.9, 0.999) and epsilon=1e-08.
• Learning Rate Scheduler: Cosine type with 0.03 warmup steps.
• Epochs: Trained for 1 epoch.
• Frameworks: Developed using Transformers 5.7.0, Pytorch 2.13.0+cu130, Datasets 4.0.0, and Tokenizers 0.22.2.

Potential Use Cases

Given its specialized fine-tuning on medical data, this model is likely suitable for applications requiring nuanced understanding and generation of medical text, such as:

• Medical information extraction.
• Clinical note summarization.
• Answering medical queries.
• Supporting medical research and documentation.