4bit/vicuna-13B-1.1-GPTQ-4bit-128g

Vicuna 13B 1.1 GPTQ 4bit 128g

This is a 4-bit GPTQ version of the Vicuna 13B 1.1 model.

It was created by merging the deltas provided in the above repo with the original Llama 13B model, using the code provided on their Github page.

It was then quantized to 4bit using GPTQ-for-LLaMa.

Want to try this in Colab for free?

Check out this Google Colab provided by eucdee: Google Colab for Vicuna 1.1

My Vicuna 1.1 model repositories

I have the following Vicuna 1.1 repositories available:

13B models:

• Unquantized 13B 1.1 model for GPU - HF format
• GPTQ quantized 4bit 13B 1.1 for GPU - safetensors and pt formats
• GPTQ quantized 4bit 13B 1.1 for CPU - GGML format for llama.cpp

7B models:

• Unquantized 7B 1.1 model for GPU - HF format
• GPTQ quantized 4bit 7B 1.1 for GPU - safetensors and pt formats
• GPTQ quantized 4bit 7B 1.1 for CPU - GGML format for llama.cpp

GIBBERISH OUTPUT

If you get gibberish output, it is because you are using the safetensors file without updating GPTQ-for-LLaMA.

If you use the safetensors file you must have the latest version of GPTQ-for-LLaMA inside text-generation-webui.

If you don't want to update, or you can't, use the pt file instead.

Either way, please read the instructions below carefully.

Provided files

Two model files are provided. Ideally use the safetensors file. Full details below:

Details of the files provided:

• vicuna-13B-1.1-GPTQ-4bit-128g.safetensors
  • safetensors format, with improved file security, created with the latest GPTQ-for-LLaMa code.
  • Command to create:
    • python3 llama.py vicuna-13B-1.1-HF c4 --wbits 4 --true-sequential --act-order --groupsize 128 --save_safetensors vicuna-13B-1.1-GPTQ-4bit-128g.safetensors
• vicuna-13B-1.1-GPTQ-4bit-128g.no-act-order.pt
  • pt format file, created without the --act-order flag.
  • This file may have slightly lower quality, but is included as it can be used without needing to compile the latest GPTQ-for-LLaMa code.
  • It should hopefully therefore work with one-click-installers on Windows, which include the older GPTQ-for-LLaMa code.
  • Command to create:
    • python3 llama.py vicuna-13B-1.1-HF c4 --wbits 4 --true-sequential --groupsize 128 --save_safetensors vicuna-13B-1.1-GPTQ-4bit-128g.no-act-order.pt

How to run in text-generation-webui

File vicuna-13B-1.1-GPTQ-4bit-128g.no-act-order.pt can be loaded the same as any other GPTQ file, without requiring any updates to oobaboogas text-generation-webui.

The safetensors model file was created with the latest GPTQ code, and uses --act-order to give the maximum possible quantisation quality, but this means it requires that the latest GPTQ-for-LLaMa is used inside the UI.

If you want to use the safetensors file and need to update GPTQ-for-LLaMa, here are the commands I used to clone the Triton branch of GPTQ-for-LLaMa, clone text-generation-webui, and install GPTQ into the UI:

# We need to clone GPTQ-for-LLaMa as of April 13th, due to breaking changes in more recent commits
git clone -n  https://github.com/qwopqwop200/GPTQ-for-LLaMa gptq-safe
cd gptq-safe && git checkout 58c8ab4c7aaccc50f507fd08cce941976affe5e0

# Now clone text-generation-webui, if you don't already have it
git clone https://github.com/oobabooga/text-generation-webui
# And link GPTQ-for-Llama into text-generation-webui
mkdir -p text-generation-webui/repositories
ln -s gptq-safe text-generation-webui/repositories/GPTQ-for-LLaMa

Then install this model into text-generation-webui/models and launch the UI as follows:

cd text-generation-webui
python server.py --model vicuna-13B-1.1-GPTQ-4bit-128g --wbits 4 --groupsize 128 --model_type Llama # add any other command line args you want

The above commands assume you have installed all dependencies for GPTQ-for-LLaMa and text-generation-webui. Please see their respective repositories for further information.

If you are on Windows, or cannot use the Triton branch of GPTQ for any other reason, you can instead use the CUDA branch:

git clone https://github.com/qwopqwop200/GPTQ-for-LLaMa -b cuda
cd GPTQ-for-LLaMa
python setup_cuda.py install

Then link that into text-generation-webui/repositories as described above.

Or just use vicuna-13B-1.1-GPTQ-4bit-128g.no-act-order.pt as mentioned above, which should work without any upgrades to text-generation-webui.

Vicuna Model Card

Model details

Model type: Vicuna is an open-source chatbot trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT. It is an auto-regressive language model, based on the transformer architecture.

Model date: Vicuna was trained between March 2023 and April 2023.

Organizations developing the model: The Vicuna team with members from UC Berkeley, CMU, Stanford, and UC San Diego.

Paper or resources for more information: https://vicuna.lmsys.org/

License: Apache License 2.0

Where to send questions or comments about the model: https://github.com/lm-sys/FastChat/issues

Intended use

Primary intended uses: The primary use of Vicuna is research on large language models and chatbots.

Primary intended users: The primary intended users of the model are researchers and hobbyists in natural language processing, machine learning, and artificial intelligence.

Training dataset

70K conversations collected from ShareGPT.com.

Evaluation dataset

A preliminary evaluation of the model quality is conducted by creating a set of 80 diverse questions and utilizing GPT-4 to judge the model outputs. See https://vicuna.lmsys.org/ for more details.

Major updates of weights v1.1

• Refactor the tokenization and separator. In Vicuna v1.1, the separator has been changed from "###" to the EOS token "</s>". This change makes it easier to determine the generation stop criteria and enables better compatibility with other libraries.
• Fix the supervised fine-tuning loss computation for better model quality.