OpenCM-14 / README.md
leaderboard-pr-bot's picture
Adding Evaluation Results
9b990a5 verified
metadata
license: cc-by-nc-4.0
tags:
  - merge
model-index:
  - name: OpenCM-14
    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: 69.28
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cookinai/OpenCM-14
          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: 86.89
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cookinai/OpenCM-14
          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: 65.01
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cookinai/OpenCM-14
          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: 61.07
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cookinai/OpenCM-14
          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: 81.29
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cookinai/OpenCM-14
          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: 72.93
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cookinai/OpenCM-14
          name: Open LLM Leaderboard

Finetune of cookinai/CM-14 with the teknium/openhermes dataset. My first finetune, might have some bugs/overfitting, might reupload

Previous model had stopping token errors causing issues with the final token in the ChatML preset. This finetuning job should fix any prompt template errors. Please tell me if you get any such errors.

Heard that this error is common amongst heavily merged macaroni models. Might try to stray away from them in the future or dilute them with other models.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric Value
Avg. 72.75
AI2 Reasoning Challenge (25-Shot) 69.28
HellaSwag (10-Shot) 86.89
MMLU (5-Shot) 65.01
TruthfulQA (0-shot) 61.07
Winogrande (5-shot) 81.29
GSM8k (5-shot) 72.93