README.md

metadata

language:
  - en
license: cc-by-nc-nd-4.0
tags:
  - UNA
  - SOLAR
  - MathPILE
datasets:
  - GAIR/MathPile
model-index:
  - name: UNA-POLAR-10.7B-InstructMath-v2
    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: 70.73
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fblgit/UNA-POLAR-10.7B-InstructMath-v2
          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: 88.2
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fblgit/UNA-POLAR-10.7B-InstructMath-v2
          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: 66.03
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fblgit/UNA-POLAR-10.7B-InstructMath-v2
          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: 71.73
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fblgit/UNA-POLAR-10.7B-InstructMath-v2
          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: 82.95
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fblgit/UNA-POLAR-10.7B-InstructMath-v2
          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: 64.75
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fblgit/UNA-POLAR-10.7B-InstructMath-v2
          name: Open LLM Leaderboard

UNA-POLAR-10.7B-InstructMath-v2

Model description

Its a UNA version with DPO over MathPILE Books out of the UNA-SOLAR-10.7B-Instruct-1.0

I used MathPILE OUTSTANDING Dataset of great Mathematic material in order to produce this beautiful model :)

Intended uses & limitations

If your model has inside UNA technology, cite.

Training and evaluation data

UNA-DPO over Attention and MLP's

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2-UNA
• Pytorch 2.1.2+cu121
• Datasets 2.16.0
• Tokenizers 0.15.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	74.07
AI2 Reasoning Challenge (25-Shot)	70.73
HellaSwag (10-Shot)	88.20
MMLU (5-Shot)	66.03
TruthfulQA (0-shot)	71.73
Winogrande (5-shot)	82.95
GSM8k (5-shot)	64.75