README.md

metadata

license: apache-2.0
model-index:
  - name: Mistral-T5-7B-v1
    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: 68.6
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ignos/Mistral-T5-7B-v1
          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: 86.3
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ignos/Mistral-T5-7B-v1
          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: 64.62
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ignos/Mistral-T5-7B-v1
          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: 61.86
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ignos/Mistral-T5-7B-v1
          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: 80.27
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ignos/Mistral-T5-7B-v1
          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: 73.16
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ignos/Mistral-T5-7B-v1
          name: Open LLM Leaderboard

Model Card for Model ID

This model is a finetuning of Toten5/Marcoroni-neural-chat-7B-v2

Model Details

Model Description

• Developed by: Ignos
• Model type: Mistral
• License: Apache-2.0

Uses

Model created to improve instructional behavior.

Bias, Risks, and Limitations

The same bias, risks and limitations from base models.

Training Details

Training Data

• tatsu-lab/alpaca

Training Procedure

• Training with QLoRA approach and merging with base model.

Results

• Huggingface evaluation pending

Summary

Technical Specifications

Model Architecture and Objective

• Models based on Mistral Architecture

Compute Infrastructure

• Training on RunPod

Hardware

• 3 x RTX 4090
• 48 vCPU 377 GB RAM

Software

• Axolotl 0.3.0

Framework versions

• PEFT 0.6.0

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	72.47
AI2 Reasoning Challenge (25-Shot)	68.60
HellaSwag (10-Shot)	86.30
MMLU (5-Shot)	64.62
TruthfulQA (0-shot)	61.86
Winogrande (5-shot)	80.27
GSM8k (5-shot)	73.16