metadata

license: cc-by-nc-4.0
tags:
  - merge
model-index:
  - name: Lonepino-11B
    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: 68.26
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=beberik/Lonepino-11B
          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: 84.57
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=beberik/Lonepino-11B
          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: 63.76
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=beberik/Lonepino-11B
          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: 63.45
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=beberik/Lonepino-11B
          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: 78.93
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=beberik/Lonepino-11B
          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: 61.64
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=beberik/Lonepino-11B
          name: Open LLM Leaderboard

Description

This repo contains bf16 files of Lonepino-11B. Just a normal model.

Model used

The secret sauce

neural-maid-11B

slices:
  - sources:
    - model: Intel/neural-chat-7b-v3-3-Slerp
      layer_range: [0, 24]
  - sources:
    - model: NeverSleep/Noromaid-7b-v0.2
      layer_range: [8, 32]

merge_method: passthrough
dtype: bfloat16

loyal-PiVoT-11B

slices:
  - sources:
    - model: chargoddard/loyal-piano-m7-cdpo
      layer_range: [0, 24]
  - sources:
    - model: maywell/PiVoT-0.1-Starling-LM-RP
      layer_range: [8, 32]

merge_method: passthrough
dtype: bfloat16

Lonepino-11B :

slices:
  - sources:
      - model: "./neural-maid-11B"
        layer_range: [0, 48]
      - model: "./loyal-PiVoT-11B"
        layer_range: [0, 48]
merge_method: slerp
base_model: "./neural-maid-11B"
parameters:
  t:
    - value: 0.4
dtype: bfloat16

Prompt template

Alpaca. Or chatml. Or any you like.

=w=

I use mergekit for all the manipulation told here.

Thanks to the Undi95 for the original 11B mistral merge recipe.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	70.10
AI2 Reasoning Challenge (25-Shot)	68.26
HellaSwag (10-Shot)	84.57
MMLU (5-Shot)	63.76
TruthfulQA (0-shot)	63.45
Winogrande (5-shot)	78.93
GSM8k (5-shot)	61.64