README.md

metadata

license: apache-2.0
tags:
  - merge
model-index:
  - name: Slerp-CM-mist-dpo
    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: 69.62
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Slerp-CM-mist-dpo
          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: 87.09
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Slerp-CM-mist-dpo
          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: 64.81
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Slerp-CM-mist-dpo
          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: 62.82
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Slerp-CM-mist-dpo
          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: 81.45
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Slerp-CM-mist-dpo
          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: 72.78
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=abacusai/Slerp-CM-mist-dpo
          name: Open LLM Leaderboard

This model is a Slerp Merge of cookinai/CatMacaroni-Slerp and mncai/mistral-7b-dpo-v5.

Evaluation Results

HuggingFace Leaderboard

Average	ARC	HellaSwag	MMLU	TruthfulQA	Winogrande	GSM8K
73.1	69.62	87.09	64.81	62.82	81.45	72.78

The model did achieve an improvement in TruthfulQA over cookinai/CatMacaroni-Slerp and GSM8K over mncai/mistral-7b-dpo-v5 which was the goal of the merge leading to an average score that was a better than both. It is unclear why the TruthfulQA metric is still meaningfully lower than the base mncai/mistral-7b-dpo-v5.

Training Details

.yaml file for mergekit

slices:
  - sources:
      - model: cookinai/CatMacaroni-Slerp
        layer_range: [0, 32]
      - model: mncai/mistral-7b-dpo-v5
        layer_range: [0, 32]
merge_method: slerp
base_model: mncai/mistral-7b-dpo-v5
parameters:
  t:
    - filter: self_attn
      value: [0, 0.5, 0.3, 0.7, 1]
    - filter: mlp
      value: [1, 0.5, 0.7, 0.3, 0]
    - value: 0.5 # fallback for rest of tensors
dtype: float16

Bias, Risks, and Limitations

The model has not been evaluated for safety and is only intended for research and experiments.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	73.10
AI2 Reasoning Challenge (25-Shot)	69.62
HellaSwag (10-Shot)	87.09
MMLU (5-Shot)	64.81
TruthfulQA (0-shot)	62.82
Winogrande (5-shot)	81.45
GSM8k (5-shot)	72.78