rwitz_-_go-bruins-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

go-bruins - GGUF

• Model creator: https://huggingface.co/rwitz/
• Original model: https://huggingface.co/rwitz/go-bruins/

Original model description:

language:

• en license: cc-by-nc-4.0 base_model: Q-bert/MetaMath-Cybertron-Starling datasets:
• Intel/orca_dpo_pairs pipeline_tag: text-generation model-index:
• name: go-bruins 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: 69.11 name: normalized accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=rwitz/go-bruins 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: 86.73 name: normalized accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=rwitz/go-bruins 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: 64.94 name: accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=rwitz/go-bruins 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: 58.71 source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=rwitz/go-bruins 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: 81.45 name: accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=rwitz/go-bruins 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: 69.9 name: accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=rwitz/go-bruins name: Open LLM Leaderboard

---

Go Bruins - A Fine-tuned Language Model

Join my AI Discord: rwitz

Updates

December 9, 2023: Go-Bruins has placed #6 overall and #1 for 7 billion parameter models on the Hugging Face Leaderboard!

Overview

Go Bruins is a state-of-the-art language model fine-tuned on the Q-bert/MetaMath-Cybertron-Starling architecture. It's designed to push the boundaries of NLP applications, offering unparalleled performance in generating human-like text.

Model Details

• Developer: Ryan Witzman
• Base Model: Q-bert/MetaMath-Cybertron-Starling
• Fine-tuning Method: Direct Preference Optimization (DPO)
• Training Steps: 200
• Language: English
• License: MIT

Capabilities

Go Bruins excels in a variety of NLP tasks, including but not limited to:

• Text generation
• Language understanding
• Sentiment analysis

Usage

Warning: This model may output NSFW or illegal content. Use with caution and at your own risk.

For Direct Use:

from transformers import pipeline

model_name = "rwitz/go-bruins"
inference_pipeline = pipeline('text-generation', model=model_name)

input_text = "Your input text goes here"
output = inference_pipeline(input_text)

print(output)

GGUF Quantized Files are Located at NyxKrage/go-bruins-GGUF

Not Recommended For:

• Illegal activities
• Harassment
• Professional advice or crisis situations

Training and Evaluation

Trained on a dataset from Intel/orca_dpo_pairs, Go Bruins has shown promising improvements over its predecessor, Q-Bert.

Evaluations

Go-Bruins is the SOTA 7B model.

Note: The original MMLU evaluation has been corrected to include 5-shot data rather than 1-shot data.

Contact

For any inquiries or feedback, reach out to Ryan Witzman on Discord: rwitz_.

---

Citations

@misc{unacybertron7b,
  title={Cybertron: Uniform Neural Alignment}, 
  author={Xavier Murias},
  year={2023},
  publisher = {HuggingFace},
  journal = {HuggingFace repository},
  howpublished = {\url{https://huggingface.co/fblgit/una-cybertron-7b-v2-bf16}},
}

This model card was created with care by Ryan Witzman.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here