1rsh's picture
Update README.md
9e7d99b verified
metadata
base_model: unsloth/llama-3.2-3b-bnb-4bit
tags:
  - text-generation-inference
  - transformers
  - unsloth
  - llama
  - trl
  - synthetic-data
license: apache-2.0
language:
  - en
datasets:
  - 1rsh/SPInO-ParkinsonsDiseaseQA
widget:
  - text: What is the effect of alpha-synuclein on microglia?
model-index:
  - name: Llama 3.2 3B | Parkinson's Disease
    results:
      - task:
          type: question-answering
          name: Question Answering
        dataset:
          name: 1rsh/SPInO-ParkinsonsDiseaseQA
          type: 1rsh/SPInO-ParkinsonsDiseaseQA
          split: test
        metrics:
          - type: rouge
            value: 0.432
            name: ROUGE-1
      - task:
          type: question-answering
          name: Question Answering
        dataset:
          name: 1rsh/SPInO-ParkinsonsDiseaseQA
          type: 1rsh/SPInO-ParkinsonsDiseaseQA
          split: test
        metrics:
          - type: rouge
            value: 0.253
            name: ROUGE-2
      - task:
          type: question-answering
          name: Question Answering
        dataset:
          name: 1rsh/SPInO-ParkinsonsDiseaseQA
          type: 1rsh/SPInO-ParkinsonsDiseaseQA
          split: test
        metrics:
          - type: rouge
            value: 0.372
            name: ROUGE-L
      - task:
          type: question-answering
          name: Question Answering
        dataset:
          name: 1rsh/SPInO-ParkinsonsDiseaseQA
          type: 1rsh/SPInO-ParkinsonsDiseaseQA
          split: test
        metrics:
          - type: meteor
            value: 0.386
            name: METEOR
pipeline_tag: question-answering

Uploaded model

  • Developed by: 1rsh
  • License: apache-2.0
  • Finetuned from model : unsloth/llama-3.2-3b-bnb-4bit
  • Finetuned using: 1rsh/SPInO-ParkinsonsDiseaseQA

This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.

This model has significantly better performance over base unsloth/Llama-3.2-3B!

relative_performance =
{
  'ROUGE-1': 1.510526033477611,
  'ROUGE-2': 1.9383727599984661,
  'ROUGE-L': 1.6544738302783082,
  'METEOR': 1.1253973831879713
}