leaderboard-pr-bot's picture
Adding Evaluation Results
2a11eb0 verified
|
raw
history blame
5.62 kB
metadata
license: llama2
model-index:
  - name: speechless-mistral-7b-dare-0.85
    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: 63.31
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=speechlessai/speechless-mistral-7b-dare-0.85
          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: 84.93
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=speechlessai/speechless-mistral-7b-dare-0.85
          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: 64.22
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=speechlessai/speechless-mistral-7b-dare-0.85
          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: 50.68
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=speechlessai/speechless-mistral-7b-dare-0.85
          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: 79.32
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=speechlessai/speechless-mistral-7b-dare-0.85
          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: 19.86
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=speechlessai/speechless-mistral-7b-dare-0.85
          name: Open LLM Leaderboard

Experiment for DARE(Drop and REscale), most of the delta parameters can be directly set to zeros without affecting the capabilities of SFT LMs and larger models can tolerate a higher proportion of discarded parameters.

Merged with below DARE models.

weight_mask_rate: 0.85 / use_weight_rescale: True / mask_stratery: random / scaling_coefficient: 1.0

Model Average ARC HellaSwag MMLU TruthfulQA Winogrande GSM8K
Intel/neural-chat-7b-v3-1 61.59 66.21 83.64 62.37 59.65 78.14 19.56
migtissera/SynthIA-7B-v1.3 59.34 62.12 83.45 62.65 51.37 78.85 17.59
bhenrym14/mistral-7b-platypus-fp16 58.71 63.05 84.15 64.11 45.07 78.53 17.36
jondurbin/airoboros-m-7b-3.1.2 58.75 61.86 83.51 61.91 53.75 77.58 13.87
teknium/CollectiveCognition-v1.1-Mistral-7B 62.92 62.12 84.17 62.35 57.62 75.37 15.62
uukuguy/speechless-mistral-dolphin-orca-platypus-samantha-7b 62.06 64.33 84.4 63.72 52.52 78.37 21.38
speechless-mistral-7b-dare-0.85 (Merge 6 DARE models) 64.69 63.57 84.82 64.29 50.66 79.24 45.56

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric Value
Avg. 60.39
AI2 Reasoning Challenge (25-Shot) 63.31
HellaSwag (10-Shot) 84.93
MMLU (5-Shot) 64.22
TruthfulQA (0-shot) 50.68
Winogrande (5-shot) 79.32
GSM8k (5-shot) 19.86