README.md

---
language:
- en
dataset_info:
  features:
  - name: question
    dtype: string
  - name: answer
    dtype: string
  - name: instruction
    dtype: string
  splits:
  - name: train
    num_bytes: 4328618
    num_examples: 3000
  download_size: 1750950
  dataset_size: 4328618
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---
# Dataset Card for "medical_biological_instruction_format"

Best advice:
- drop the instruction column or use your choice of high-functioning llm to generate variant prompts
- for medical and science database expansions, i'm having excellent results using:
  - [SciPhi/SciPhi-Mistral-7B-32k](https://huggingface.co/SciPhi/SciPhi-Mistral-7B-32k)
  - [TheBloke/SciPhi-Mistral-7B-32k-GGUF](https://huggingface.co/TheBloke/SciPhi-Mistral-7B-32k-GGUF)
  - Or if you have RAG setup on a corpus:
    - [SciPhi/Sensei-7B-V2](https://huggingface.co/SciPhi/Sensei-7B-V2)
    - [TheBloke/Sensei-7B-V1-GGUF](https://huggingface.co/TheBloke/Sensei-7B-V1-GGUF)
  - [Falconsai/medical_summarization](https://huggingface.co/Falconsai/medical_summarization)