README.md

metadata

license: cc-by-nc-4.0
tags:
  - merge
  - mergekit
datasets:
  - pankajmathur/orca_mini_v1_dataset
  - openai/summarize_from_feedback
  - PygmalionAI/PIPPA
  - chargoddard/rpguild
  - lemonilia/LimaRP
  - PKU-Alignment/PKU-SafeRLHF
  - Intel/orca_dpo_pairs
  - allenai/ultrafeedback_binarized_cleaned
model-index:
  - name: piano-medley-7b
    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: 67.58
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=chargoddard/piano-medley-7b
          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: 85.36
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=chargoddard/piano-medley-7b
          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: 64.49
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=chargoddard/piano-medley-7b
          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: 61.42
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=chargoddard/piano-medley-7b
          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: 79.16
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=chargoddard/piano-medley-7b
          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: 56.56
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=chargoddard/piano-medley-7b
          name: Open LLM Leaderboard

Another experiment in the line of loyal-piano-m7.

Steps taken to produce this model:

• Train loyal-piano-m7
• cDPO with HuggingFaceH4/ultrafeedback_binarized to produce loyal-piano-m7-cdpo
• Train another model with different sampling of the same source datasets as loyal-piano, let's call it servile-harpsichord
• cDPO servile-harpsichord with allenai/ultrafeedback_binarized_cleaned, Intel/orca_dpo_pairs, and a helpfulness-only version of PKU-Alignment/PKU-SafeRLHF
• TIES merge several checkpoints of servile-harpsichord-cdpo with loyal-piano-m7-cdpo

Local benchmarks show the result to be better than any of the individual components. Let's see if that holds up!

Trained using the Alpaca prompt format.

Configuration for final merge:

models:
  - model: chargoddard/loyal-piano-m7-cdpo
    parameters:
      density: 1.0
      weight: 1.0
  - model: /home/ubuntu/servile-harpsichord-cdpo/checkpoint-4186
    parameters:
      weight: 0.1
  - model: /home/ubuntu/servile-harpsichord-cdpo/checkpoint-5796
    parameters:
      weight: 0.2
  - model: /home/ubuntu/servile-harpsichord-cdpo/checkpoint-6118
    parameters:
      weight: 0.3
  - model: /home/ubuntu/servile-harpsichord-cdpo/final
    parameters:
      weight: 0.4
merge_method: ties
base_model: mistralai/Mistral-7B-v0.1
dtype: bfloat16
parameters:
  density: 0.4
  normalize: true
  int8_mask: true

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	69.10
AI2 Reasoning Challenge (25-Shot)	67.58
HellaSwag (10-Shot)	85.36
MMLU (5-Shot)	64.49
TruthfulQA (0-shot)	61.42
Winogrande (5-shot)	79.16
GSM8k (5-shot)	56.56