diffnamehard/Psyfighter2-Noromaid-ties-Capybara-13B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:13BQuant:FP8Ctx Length:4kPublished:Jan 10, 2024License:cc-by-nc-4.0Architecture:Transformer0.0K Open Weights Warm

diffnamehard/Psyfighter2-Noromaid-ties-Capybara-13B is a 13 billion parameter language model developed by diffnamehard, fine-tuned on the LDJnr/Capybara dataset. This model is based on the Psyfighter2-Noromaid-ties-13B architecture and is intended for experimental purposes. It demonstrates general language understanding capabilities across various benchmarks, including ARC, HellaSwag, and MMLU.

Loading preview...

Model Overview

diffnamehard/Psyfighter2-Noromaid-ties-Capybara-13B is an experimental 13 billion parameter language model. It was created by fine-tuning the existing Psyfighter2-Noromaid-ties-13B model on the LDJnr/Capybara dataset.

Key Capabilities & Performance

This model exhibits general language understanding and reasoning abilities, as indicated by its performance across several benchmarks:

  • ARC (25-shot): 62.29
  • HellaSwag (10-shot): 83.87
  • MMLU (5-shot): 56.59
  • TruthfulQA (0-shot): 51.44
  • Winogrande (5-shot): 77.03
  • GSM8K (5-shot): 30.4

The average score across these benchmarks is 60.27.

Intended Use

This model is primarily for experimental purposes and is not described as stable for production environments. Developers interested in exploring models fine-tuned on the Capybara dataset or evaluating the performance characteristics of the Psyfighter2-Noromaid-ties architecture may find this model useful for research and development.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p