README.md

metadata

language:
  - en
license: apache-2.0
tags:
  - wikihow
  - tutorial
  - educational
datasets:
  - ajibawa-2023/WikiHow
model-index:
  - name: WikiHow-Mistral-Instruct-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: 60.92
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ajibawa-2023/WikiHow-Mistral-Instruct-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: 80.99
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ajibawa-2023/WikiHow-Mistral-Instruct-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: 58.57
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ajibawa-2023/WikiHow-Mistral-Instruct-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: 62.16
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ajibawa-2023/WikiHow-Mistral-Instruct-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: 74.82
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ajibawa-2023/WikiHow-Mistral-Instruct-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: 30.02
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ajibawa-2023/WikiHow-Mistral-Instruct-7B
          name: Open LLM Leaderboard

WikiHow-Mistral-Instruct-7B

This Model is trained on my WikiHow dataset.

This model is very very good with generating tutorials in the style of WikiHow. By leveraging this repository of practical knowledge, the model has been trained to comprehend and generate text that is highly informative and instructional in nature. The depth and accuracy of generated tutorials is exceptional. The WikiHow dataset encompasses a wide array of topics, ranging from everyday tasks to specialized skills, making it an invaluable resource for refining the capabilities of language models. Through this fine-tuning process, the model has been equipped with the ability to offer insightful guidance and assistance across diverse domains.

This is a fully finetuned model. Links for Quantized models are given below.

GGUF & Exllama

GGUF: Link

Exllama v2: Link

Special Thanks to Bartowski for quantizing this model.

Training:

Entire dataset was trained on 4 x A100 80GB. For 3 epoch, training took more than 15 Hours. Axolotl codebase was used for training purpose. Entire data is trained on Mistral-7B-Instruct-v0.2.

Example Prompt:

This model uses ChatML prompt format.

<|im_start|>system
You are a Helpful Assistant who can write long and very detailed tutorial on various subjects in the styles of WiKiHow. Include in depth explanations for each step and how it helps achieve the desired outcome, including key tips and guidelines.<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

You can modify above Prompt as per your requirement.

I want to say special Thanks to the Open Source community for helping & guiding me to better understand the AI/Model development.

Thank you for your love & support.

Example Output

Example 1

Example 2

Example 3

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	61.25
AI2 Reasoning Challenge (25-Shot)	60.92
HellaSwag (10-Shot)	80.99
MMLU (5-Shot)	58.57
TruthfulQA (0-shot)	62.16
Winogrande (5-shot)	74.82
GSM8k (5-shot)	30.02