README.md

metadata

license: cc-by-nc-4.0
extra_gated_prompt: >-
  You agree not to use the model for healthcare decision-making or commercial
  use
extra_gated_fields:
  I agree to use this model for non-commercial use ONLY: checkbox
  I agree not to use this model for healthcare decision-making: checkbox
language:
  - en

Clinical Camel

Model Description

Clinical Camel is an open large language model (LLM), fine-tuned on the LLaMA-2 70B architecture using QLoRA. It is tailored for the medical and clinical research, capable of processing and generating relevant content.

Review our pre-print for more details: Clinical Camel - Pre-print

Performance

Clinical Camel demonstrates competitive performance on medical benchmarks.

Table: Five-Shot Performance of Clinical Camel-70B (C70), GPT3.5, GPT4, and Med-PaLM 2 on Various Medical Datasets

Dataset	ClinicalCamel-70B	GPT3.5	GPT4	Med-PaLM 2
MMLU Anatomy	65.2	60.7	80.0	77.8
MMLU Clinical Knowledge	72.8	68.7	86.4	88.3
MMLU College Biology	81.2	72.9	93.8	94.4
MMLU College Medicine	68.2	63.6	76.3	80.9
MMLU Medical Genetics	69.0	68.0	92.0	90.0
MMLU Professional Medicine	75.0	69.8	93.8	95.2
MedMCQA	54.2	51.0	72.4	71.3
MedQA (USMLE)	60.7	53.6	81.4	79.7
PubMedQA	77.9	60.2	74.4	79.2
USMLE Sample Exam	64.3	58.5	86.6	-

Evaluation Datasets:

The performance of Clinical Camel was benchmarked across several datasets, including:

• USMLE Step 1
• USMLE Step 2
• USMLE Step 3
• MedMCQA
• MedQA USMLE

Evaluation Reproduction:

To reproduce the evaluations with lm-evaluation-harness see the 'TaskFiles' folder