carsenk/phi3.5_mini_exp_825_uncensored

TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:4kPublished:Aug 29, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

carsenk/phi3.5_mini_exp_825_uncensored is a 4 billion parameter Phi-3.5 model developed by Carsen Klock, fine-tuned from unsloth/phi-3.5-mini-instruct-bnb-4bit. This model was trained using Unsloth and Huggingface's TRL library, focusing on rapid training as a test. It is provided for experimental purposes, with GGUF files included for efficient inference.

Loading preview...

Model Overview

carsenk/phi3.5_mini_exp_825_uncensored is a 4 billion parameter Phi-3.5 model developed by Carsen Klock. It was fine-tuned from the unsloth/phi-3.5-mini-instruct-bnb-4bit base model, utilizing the Unsloth library and Huggingface's TRL for accelerated training.

Key Characteristics

  • Architecture: Based on the Phi-3.5 model family.
  • Parameter Count: 4 billion parameters.
  • Training Method: Trained 2x faster using Unsloth and Huggingface's TRL library.
  • Experimental Focus: Developed as a test, trained over 10,500 epochs on a single 4080 SUPER GPU.
  • Inference Support: Includes GGUF files for optimized inference with tools like llama.cpp.

Intended Use

This model is primarily intended for test purposes and experimentation, demonstrating rapid fine-tuning capabilities. Developers can use it to explore the performance of Phi-3.5 models fine-tuned with Unsloth, particularly for applications requiring efficient inference via GGUF formats.