metadata
license: afl-3.0
language:
- en
library_name: peft
datasets:
- EagleConsortium/MrEagle-2126
NOTE: DO NOT USE THIS–QUANTIZATION RENDERS THIS LORA-BASED MODEL UNUSABLE. USE THE HF PYTORCH MERGED MODEL OR GPTQ INSTEAD.
MrEagle-LoRA is a large language model (LLM) based on Llama2-7B-Chat created by finetuning a LoRA from a dataset of 2,126 1-turn conversations artificially generated using GPT-4. The conversations were designed to fit the tone of the Discord bot Mr. Eagle.
We recommend that you run the largest model that functions on your device. The Q4_K_M model offers a good balance of performance and speed.