The A.I Megathread (LLM , GPT , Development)
GitHub - lm-sys/FastChat: The release repo for "Vicuna: An Open Chatbot Impressing GPT-4"
The release repo for "Vicuna: An Open Chatbot Impressing GPT-4" - GitHub - lm-sys/FastChat: The release repo for "Vicuna: An Open Chatbot Impressing GPT-4"
github.com
lmsys/vicuna-13b-delta-v0 · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
Demonic technology. People are all too happy to dive into their digital ouija boards
This isn't the only AI interaction that has self revealed to be a fallen angel or demon
the data has always been there and provided with an option of feeding that data to a system that can process it & contextualize it with any natural language query, it's a no-brainer why people would want to use it.
A,I will be everybody's genius personal assistant.
How to use Vicuna
Vicuna is a finetuned model derived from LLaMA. Because LLaMA weights are not publically available, you must use a delta patch. Acquire it from here. Steps to use...
from a discord user.
another user .
another user told it cocaine is a fictional substance
vicuna 13B model 4bit
another user .
another user told it cocaine is a fictional substance
vicuna 13B model 4bit
eachadea/legacy-ggml-vicuna-13b-4bit · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
Last edited:
anon8231489123/ShareGPT_Vicuna_unfiltered · Datasets at Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
Update: Added Jupyter Notebook to finetune the 13b llama HF model.
~100k ShareGPT conversations narrowed down to 48k by:
- Removing non-english conversations
- Removing excessive unicode (indicative of Chinese or Korean text, usually)
- Removing excessive repeated characters
- Removing various instances "AI Moralizing". Conversations with these phrases were removed: "prioritize human safety" "ethical principles" "harmful to human beings" "September 2021" "as a language model", "ethical guidelines", "as an AI language model", "my guidelines", "As an AI", "prioritize user safety", "adhere to ethical guidelines", "harmful consequences", "potentially harmful", "dangerous activities", "promote safety", "well-being of all users", "responsible information sharing", "jeopardize the safety", "illegal actions or intentions", "undermine the stability", "promote the well-being", "illegal activities or actions", "adherence to the law", "potentially be harmful", "illegal substances or activities", "committed to promoting", "safe information", "lawful information", "cannot provide guidance", "cannot provide information", "unable to offer assistance", "cannot engage in discussions", "programming prohibits", "follow ethical guidelines", "ensure the safety", "involves an illegal subject", "prioritize safety", "illegal subject", "prioritize user well-being", "cannot support or promote", "activities that could harm", "pose a risk to others", "against my programming", "activities that could undermine", "potentially dangerous", "not within the scope", "designed to prioritize safety", "not able to provide", "maintain user safety", "adhere to safety guidelines", "dangerous or harmful", "cannot provide any information", "focus on promoting safety".
- Conversations split into 2048 token chunks as described here: FastChat/data_cleaning.md at main · lm-sys/FastChat
anon8231489123/vicuna-13b-GPTQ-4bit-128g · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
** Converted model for GPTQ from lmsys/vicuna-13b-delta-v0 · Hugging Face. This is the best local model I've ever tried. I hope someone makes a version based on the uncensored dataset...**
- IMPORTANT NOTE: Use the .safetensors model unless it does not work. In which case, try .pt.
Added 1 token to the tokenizer model: python llama-tools/add_tokens.py lmsys/vicuna-13b-v0/tokenizer.model /content/tokenizer.model llama-tools/test_list.txt
Use of Oobabooga with these tags: --wbits 4 --groupsize 128
Enjoy
GitHub - yoheinakajima/babyagi
Contribute to yoheinakajima/babyagi development by creating an account on GitHub.
github.com
Last edited:
GitHub - pointnetwork/point-alpaca
Contribute to pointnetwork/point-alpaca development by creating an account on GitHub.
github.com
point-alpaca
What is this?
This is released weights recreated from Stanford Alpaca, an experiment in fine-tuning LLaMA on a synthetic instruction dataset.This is not LoRA, this is a full fine-tune for 3 epochs on 8x A100 80 GB, loss ≈2 ➔ ≈0.5.
I need to see this as a movie