README.md

metadata

base_model:
  - FreedomIntelligence/HuatuoGPT-o1-8B
  - Skywork/Skywork-o1-Open-Llama-3.1-8B
library_name: transformers
pipeline_tag: text-generation
tags:
  - mergekit
  - merge
license: llama3.1
model-index:
  - name: HuatuoSkywork-o1-Llama-3.1-8B
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: IFEval (0-Shot)
          type: wis-k/instruction-following-eval
          split: train
          args:
            num_few_shot: 0
        metrics:
          - type: inst_level_strict_acc and prompt_level_strict_acc
            value: 39.61
            name: averaged accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=grimjim%2FHuatuoSkywork-o1-Llama-3.1-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: BBH (3-Shot)
          type: SaylorTwift/bbh
          split: test
          args:
            num_few_shot: 3
        metrics:
          - type: acc_norm
            value: 28.33
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=grimjim%2FHuatuoSkywork-o1-Llama-3.1-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MATH Lvl 5 (4-Shot)
          type: lighteval/MATH-Hard
          split: test
          args:
            num_few_shot: 4
        metrics:
          - type: exact_match
            value: 33.99
            name: exact match
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=grimjim%2FHuatuoSkywork-o1-Llama-3.1-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: GPQA (0-shot)
          type: Idavidrein/gpqa
          split: train
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 5.7
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=grimjim%2FHuatuoSkywork-o1-Llama-3.1-8B
          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: 11.12
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=grimjim%2FHuatuoSkywork-o1-Llama-3.1-8B
          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: 23.28
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=grimjim%2FHuatuoSkywork-o1-Llama-3.1-8B
          name: Open LLM Leaderboard

HuatuoSkywork-o1-Llama-3.1-8B

This is a merge of pre-trained language models created using mergekit.

This is an experiment to see what happens when two o1-inspired models are merged. The result achieves an unexpectedly high MATH Lvl 5 benchmark of 33.99%.

This model is capable of roleplay text completion, but it will tend to drive narrative along chain-of-thought lines. Technically, an assistant persona is a role-play, so this model could be an interesting or good fit, depending on one's taste.

Built with Llama.

Merge Details

Merge Method

This model was merged using the SLERP merge method.

Models Merged

The following models were included in the merge:

• FreedomIntelligence/HuatuoGPT-o1-8B
• Skywork/Skywork-o1-Open-Llama-3.1-8B

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: Skywork/Skywork-o1-Open-Llama-3.1-8B
  - model: FreedomIntelligence/HuatuoGPT-o1-8B
merge_method: slerp
base_model: Skywork/Skywork-o1-Open-Llama-3.1-8B
parameters:
  t:
   - value: 0.5
dtype: bfloat16

Open LLM Leaderboard Evaluation Results

Detailed results can be found here! Summarized results can be found here!

Metric	Value (%)
Average	23.67
IFEval (0-Shot)	39.61
BBH (3-Shot)	28.33
MATH Lvl 5 (4-Shot)	33.99
GPQA (0-shot)	5.70
MuSR (0-shot)	11.12
MMLU-PRO (5-shot)	23.28