hertabot_gguf_v0.1

Uploaded model

• Developed by: Vexxations
• License: apache-2.0
• Finetuned from model : unsloth/llama-3-8b-Instruct-bnb-4bit

4bit and 8bit quantized GGUF files for Hertabot v0.1, a conversational chatbot finetuned from llama-3-8b-instruct for deploying on Discord. This is an incredibly early-stage model that will be replaced as soon as improvements are made - and there are PLENTY

Current features:

• Has the personality and text mannerisms of Herta given an appropriate system prompt
• Exclusively text-based, has /some/ ability to interpret contexts
• TO BE TESTED

Planned additions:

• Discord action integration: allow for usage of images, emojis, gifs and other features on discord
• Memory: remember users by name, traits inferred from messages
• User recognition: distinguish users in context and address them accordingly
• Enhanced persona believability: ensure Herta has no knowledge of "real life" subjects or distances herself from it (technically Earth is insignificant to her, and she's from the future)
• Improved system prompt: Prompt atm is using Deepseekslop and I'd rather look over it and finetune it myself
• Etc.

This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.

<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>