metadata

license: llama3.1
library_name: transformers
tags:
  - moe
  - frankenmoe
  - merge
  - mergekit
base_model:
  - argilla-warehouse/Llama-3.1-8B-MagPie-Ultra
  - sequelbox/Llama3.1-8B-PlumCode
  - sequelbox/Llama3.1-8B-PlumMath
  - ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.2
model-index:
  - name: L3.1-Moe-4x8B-v0.1
    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: 43.47
            name: strict accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=moeru-ai/L3.1-Moe-4x8B-v0.1
          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: 27.86
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=moeru-ai/L3.1-Moe-4x8B-v0.1
          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: 11.1
            name: exact match
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=moeru-ai/L3.1-Moe-4x8B-v0.1
          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: 1.23
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=moeru-ai/L3.1-Moe-4x8B-v0.1
          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: 3.98
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=moeru-ai/L3.1-Moe-4x8B-v0.1
          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: 27.27
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=moeru-ai/L3.1-Moe-4x8B-v0.1
          name: Open LLM Leaderboard

L3.1-Moe-4x8B-v0.1

This model is a Mixture of Experts (MoE) made with mergekit-moe. It uses the following base models:

Heavily inspired by mlabonne/Beyonder-4x7B-v3.

Quantized models

GGUF by mradermacher

Configuration

base_model: argilla-warehouse/Llama-3.1-8B-MagPie-Ultra
gate_mode: hidden
dtype: bfloat16
experts:
  - source_model: argilla-warehouse/Llama-3.1-8B-MagPie-Ultra
    positive_prompts:
      - "chat"
      - "assistant"
      - "tell me"
      - "explain"
      - "I want"
  - source_model: sequelbox/Llama3.1-8B-PlumCode
    positive_prompts:
      - "code"
      - "python"
      - "javascript"
      - "programming"
      - "algorithm"
  - source_model: sequelbox/Llama3.1-8B-PlumMath
    positive_prompts:
      - "reason"
      - "math"
      - "mathematics"
      - "solve"
      - "count"
  - source_model: ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.2
    positive_prompts:
      - "storywriting"
      - "write"
      - "scene"
      - "story"
      - "character"

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	19.15
IFEval (0-Shot)	43.47
BBH (3-Shot)	27.86
MATH Lvl 5 (4-Shot)	11.10
GPQA (0-shot)	1.23
MuSR (0-shot)	3.98
MMLU-PRO (5-shot)	27.27