README.md

metadata

license: mit
base_model:
  - meta-llama/Llama-2-7b
datasets:
  - tatsu-lab/alpaca
model-index:
  - name: Llama2-7B-SimPO
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: IFEval (0-Shot)
          type: HuggingFaceH4/ifeval
          args:
            num_few_shot: 0
        metrics:
          - type: inst_level_strict_acc and prompt_level_strict_acc
            value: 16.59
            name: strict accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=sabersaleh/Llama2-7B-SimPO
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: BBH (3-Shot)
          type: BBH
          args:
            num_few_shot: 3
        metrics:
          - type: acc_norm
            value: 8.98
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=sabersaleh/Llama2-7B-SimPO
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MATH Lvl 5 (4-Shot)
          type: hendrycks/competition_math
          args:
            num_few_shot: 4
        metrics:
          - type: exact_match
            value: 0.76
            name: exact match
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=sabersaleh/Llama2-7B-SimPO
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: GPQA (0-shot)
          type: Idavidrein/gpqa
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 2.8
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=sabersaleh/Llama2-7B-SimPO
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MuSR (0-shot)
          type: TAUR-Lab/MuSR
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 8.59
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=sabersaleh/Llama2-7B-SimPO
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MMLU-PRO (5-shot)
          type: TIGER-Lab/MMLU-Pro
          config: main
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 7.13
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=sabersaleh/Llama2-7B-SimPO
          name: Open LLM Leaderboard

This model is aligned using the AlpacaFarm dataset, fine-tuned through the Simple Preference Optimization (SimPO) loss. The alignment process started from the Supervised Fine-Tuned (SFT) version of LLaMA 2 7B. The optimization process was conducted with a single epoch. For more information on the dataset, refer to the AlpacaFarm documentation (https://github.com/tatsu-lab/alpaca_farm).

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	7.47
IFEval (0-Shot)	16.59
BBH (3-Shot)	8.98
MATH Lvl 5 (4-Shot)	0.76
GPQA (0-shot)	2.80
MuSR (0-shot)	8.59
MMLU-PRO (5-shot)	7.13