Llama-Phi-3_DoRA / README.md
leaderboard-pr-bot's picture
Adding Evaluation Results
86e427a verified
|
raw
history blame
5.06 kB
metadata
language:
  - en
license: mit
datasets:
  - Sao10K/Claude-3-Opus-Instruct-15K
  - abacusai/SystemChat-1.1
  - Ba2han/DollyLlama-5k
model-index:
  - name: Llama-Phi-3_DoRA
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: AI2 Reasoning Challenge (25-Shot)
          type: ai2_arc
          config: ARC-Challenge
          split: test
          args:
            num_few_shot: 25
        metrics:
          - type: acc_norm
            value: 62.29
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Ba2han/Llama-Phi-3_DoRA
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: HellaSwag (10-Shot)
          type: hellaswag
          split: validation
          args:
            num_few_shot: 10
        metrics:
          - type: acc_norm
            value: 79.08
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Ba2han/Llama-Phi-3_DoRA
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MMLU (5-Shot)
          type: cais/mmlu
          config: all
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 69.44
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Ba2han/Llama-Phi-3_DoRA
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: TruthfulQA (0-shot)
          type: truthful_qa
          config: multiple_choice
          split: validation
          args:
            num_few_shot: 0
        metrics:
          - type: mc2
            value: 54.08
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Ba2han/Llama-Phi-3_DoRA
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: Winogrande (5-shot)
          type: winogrande
          config: winogrande_xl
          split: validation
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 73.4
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Ba2han/Llama-Phi-3_DoRA
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: GSM8k (5-shot)
          type: gsm8k
          config: main
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 68.01
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Ba2han/Llama-Phi-3_DoRA
          name: Open LLM Leaderboard

hf-causal-experimental (pretrained='E:/text-generation-webui/models/phi-984-dora'), limit: None, provide_description: False, num_fewshot: 5, batch_size: None

Task Version Metric Value Stderr
arc_challenge 0 acc 0.5870 ± 0.0144
acc_norm 0.6067 ± 0.0143
hendrycksTest-abstract_algebra 1 acc 0.3500 ± 0.0479
acc_norm 0.3500 ± 0.0479
hendrycksTest-college_biology 1 acc 0.8264 ± 0.0317
acc_norm 0.8264 ± 0.0317
hendrycksTest-college_chemistry 1 acc 0.4900 ± 0.0502
acc_norm 0.4900 ± 0.0502
hendrycksTest-college_mathematics 1 acc 0.3900 ± 0.0490
acc_norm 0.3900 ± 0.0490
hendrycksTest-college_physics 1 acc 0.4020 ± 0.0488
acc_norm 0.4020 ± 0.0488
winogrande 0 acc 0.7309 ± 0.0125

We have Llama-3 at home!

The model has been trained on filtered versions of tagged datasets, as well as a few thousand more examples generated with llama-3-70B.

Use Zephyr template with any system message. Default system message should be:

You are a smart, friendly and helpful assistant.

image/png

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric Value
Avg. 67.72
AI2 Reasoning Challenge (25-Shot) 62.29
HellaSwag (10-Shot) 79.08
MMLU (5-Shot) 69.44
TruthfulQA (0-shot) 54.08
Winogrande (5-shot) 73.40
GSM8k (5-shot) 68.01