TheBloke's LLM work is generously supported by a grant from andreessen horowitz (a16z)
Discolm Mixtral 8X7B v2 - GPTQ
- Model creator: Disco Research
- Original model: Discolm Mixtral 8X7B v2
WARNING - I CAN'T GET THESE GPTQ QUANTS TO WORK
Unfortunately, after 10 hours quanting at not insignificant cost, they don't actually appear to work.
I will leave them up in case any solution presents itself soon. But for now, I get errors like this
File "/workspace/venv/pytorch2/lib/python3.10/site-packages/auto_gptq/nn_modules/qlinear/qlinear_cuda_old.py", line 239, in forward
zeros = zeros.reshape(-1, 1, zeros.shape[1] * zeros.shape[2])
RuntimeError: cannot reshape tensor of 0 elements into shape [-1, 1, 0] because the unspecified dimension size -1 can be any value and is ambiguous
File "/workspace/venv/pytorch2/lib/python3.10/site-packages/auto_gptq/nn_modules/qlinear/qlinear_cuda.py", line 245, in forward
zeros = zeros.reshape(self.scales.shape)
RuntimeError: shape '[32, 8]' is invalid for input of size 0
Description
This repo contains GPTQ model files for Disco Research's Discolm Mixtral 8X7B v2.
Experimental model
This is an experimental GPTQ of MistralAI's Mixtral 7B 8Expert.
This is a quantisation of an unofficial implementation of Mixtral 7B 8Experted, created and hosted by DiscoResearch at: DiscoResearch/mixtral-7b-8expert.
To use it requires:
- Latest Transformers, installed from Github:
pip3 install git+https://github.com/huggingface/transformers.git
trust_remote_code=True
Note that I have not yet tested the model myself, I will update when I know VRAM requirements.
Multiple GPTQ parameter permutations are provided; see Provided Files below for details of the options provided, their parameters, and the software used to create them.
Repositories available
- GPTQ models for GPU inference, with multiple quantisation parameter options.
- Disco Research's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions
Prompt template: ChatML
<|im_start|>system
{system_message}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
Provided files, and GPTQ parameters
Multiple quantisation parameters are provided, to allow you to choose the best one for your hardware and requirements.
Each separate quant is in a different branch. See below for instructions on fetching from different branches.
Most GPTQ files are made with AutoGPTQ. Mistral models are currently made with Transformers.
Explanation of GPTQ parameters
- Bits: The bit size of the quantised model.
- GS: GPTQ group size. Higher numbers use less VRAM, but have lower quantisation accuracy. "None" is the lowest possible value.
- Act Order: True or False. Also known as
desc_act. True results in better quantisation accuracy. Some GPTQ clients have had issues with models that use Act Order plus Group Size, but this is generally resolved now. - Damp %: A GPTQ parameter that affects how samples are processed for quantisation. 0.01 is default, but 0.1 results in slightly better accuracy.
- GPTQ dataset: The calibration dataset used during quantisation. Using a dataset more appropriate to the model's training can improve quantisation accuracy. Note that the GPTQ calibration dataset is not the same as the dataset used to train the model - please refer to the original model repo for details of the training dataset(s).
- Sequence Length: The length of the dataset sequences used for quantisation. Ideally this is the same as the model sequence length. For some very long sequence models (16+K), a lower sequence length may have to be used. Note that a lower sequence length does not limit the sequence length of the quantised model. It only impacts the quantisation accuracy on longer inference sequences.
- ExLlama Compatibility: Whether this file can be loaded with ExLlama, which currently only supports Llama and Mistral models in 4-bit.
| Branch | Bits | GS | Act Order | Damp % | GPTQ Dataset | Seq Len | Size | ExLlama | Desc |
|---|---|---|---|---|---|---|---|---|---|
| main | 4 | None | Yes | 0.1 | VMware Open Instruct | 4096 | 4.97 GB | Yes | 4-bit, with Act Order. No group size, to lower VRAM requirements. |
| gptq-4bit-128g-actorder_True | 4 | 128 | Yes | 0.1 | VMware Open Instruct | 4096 | 5.00 GB | No | 4-bit, with Act Order and group size 128g. Uses even less VRAM than 64g, but with slightly lower accuracy. |
| gptq-4bit-32g-actorder_True | 4 | 32 | Yes | 0.1 | VMware Open Instruct | 4096 | 5.00 GB | No | 4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage. |
| gptq-3bit--1g-actorder_True | 3 | None | Yes | 0.1 | VMware Open Instruct | 4096 | 4.98 GB | No | 3-bit, with Act Order and no group size. Lowest possible VRAM requirements. May be lower quality than 3-bit 128g. |
| gptq-3bit-128g-actorder_true | 3 | 128 | Yes | 0.1 | VMware Open Instruct | 4096 | 5.00 GB | No | 3-bit, with group size 128g and act-order. Higher quality than 128g-False. |
| gptq-3bit-32g-actorder_true | 3 | 32 | Yes | 0.1 | VMware Open Instruct | 4096 | 4.99 GB | No | 3-bit, with group size 64g and act-order. Highest quality 3-bit option. |
| gptq-8bit--1g-actorder_true | 8 | None | Yes | 0.1 | VMware Open Instruct | 4096 | 4.96 GB | No | 8-bit, with Act Order. No group size, to lower VRAM requirements. |
| gptq-3bit-128g-actorder_true | 8 | 128 | Yes | 0.1 | VMware Open Instruct | 4096 | 5.00 GB | No | 8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. |
How to download, including from branches
In text-generation-webui
To download from the main branch, enter TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ in the "Download model" box.
To download from another branch, add :branchname to the end of the download name, eg TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ:gptq-4bit-128g-actorder_True
From the command line
I recommend using the huggingface-hub Python library:
pip3 install huggingface-hub
To download the main branch to a folder called DiscoLM-mixtral-8x7b-v2-GPTQ:
mkdir DiscoLM-mixtral-8x7b-v2-GPTQ
huggingface-cli download TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ --local-dir DiscoLM-mixtral-8x7b-v2-GPTQ --local-dir-use-symlinks False
To download from a different branch, add the --revision parameter:
mkdir DiscoLM-mixtral-8x7b-v2-GPTQ
huggingface-cli download TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ --revision gptq-4bit-128g-actorder_True --local-dir DiscoLM-mixtral-8x7b-v2-GPTQ --local-dir-use-symlinks False
More advanced huggingface-cli download usage
If you remove the --local-dir-use-symlinks False parameter, the files will instead be stored in the central Hugging Face cache directory (default location on Linux is: ~/.cache/huggingface), and symlinks will be added to the specified --local-dir, pointing to their real location in the cache. This allows for interrupted downloads to be resumed, and allows you to quickly clone the repo to multiple places on disk without triggering a download again. The downside, and the reason why I don't list that as the default option, is that the files are then hidden away in a cache folder and it's harder to know where your disk space is being used, and to clear it up if/when you want to remove a download model.
The cache location can be changed with the HF_HOME environment variable, and/or the --cache-dir parameter to huggingface-cli.
For more documentation on downloading with huggingface-cli, please see: HF -> Hub Python Library -> Download files -> Download from the CLI.
To accelerate downloads on fast connections (1Gbit/s or higher), install hf_transfer:
pip3 install hf_transfer
And set environment variable HF_HUB_ENABLE_HF_TRANSFER to 1:
mkdir DiscoLM-mixtral-8x7b-v2-GPTQ
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ --local-dir DiscoLM-mixtral-8x7b-v2-GPTQ --local-dir-use-symlinks False
Windows Command Line users: You can set the environment variable by running set HF_HUB_ENABLE_HF_TRANSFER=1 before the download command.
With git (not recommended)
To clone a specific branch with git, use a command like this:
git clone --single-branch --branch gptq-4bit-128g-actorder_True https://huggingface.co/TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ
Note that using Git with HF repos is strongly discouraged. It will be much slower than using huggingface-hub, and will use twice as much disk space as it has to store the model files twice (it stores every byte both in the intended target folder, and again in the .git folder as a blob.)
How to easily download and use this model in text-generation-webui
NOTE This likely doesn't work at the moment.
Please make sure you're using the latest version of text-generation-webui.
It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install.
Click the Model tab.
Under Download custom model or LoRA, enter
TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ.- To download from a specific branch, enter for example
TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ:gptq-4bit-128g-actorder_True - see Provided Files above for the list of branches for each option.
- To download from a specific branch, enter for example
Click Download.
The model will start downloading. Once it's finished it will say "Done".
In the top left, click the refresh icon next to Model.
In the Model dropdown, choose the model you just downloaded:
DiscoLM-mixtral-8x7b-v2-GPTQThe model will automatically load, and is now ready for use!
If you want any custom settings, set them and then click Save settings for this model followed by Reload the Model in the top right.
- Note that you do not need to and should not set manual GPTQ parameters any more. These are set automatically from the file
quantize_config.json.
- Note that you do not need to and should not set manual GPTQ parameters any more. These are set automatically from the file
Once you're ready, click the Text Generation tab and enter a prompt to get started!
Serving this model from Text Generation Inference (TGI)
NOTE This likely doesn't work at the moment.
It's recommended to use TGI version 1.1.0 or later. The official Docker container is: ghcr.io/huggingface/text-generation-inference:1.1.0
Example Docker parameters:
--model-id TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ --port 3000 --quantize gptq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096
Example Python code for interfacing with TGI (requires huggingface-hub 0.17.0 or later):
pip3 install huggingface-hub
from huggingface_hub import InferenceClient
endpoint_url = "https://your-endpoint-url-here"
prompt = "Tell me about AI"
prompt_template=f'''<|im_start|>system
{system_message}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
'''
client = InferenceClient(endpoint_url)
response = client.text_generation(prompt,
max_new_tokens=128,
do_sample=True,
temperature=0.7,
top_p=0.95,
top_k=40,
repetition_penalty=1.1)
print(f"Model output: {response}")
Python code example: inference from this GPTQ model
NOTE I can't get this working yet.
Install the necessary packages
Requires: Transformers 4.33.0 or later, Optimum 1.12.0 or later, and AutoGPTQ 0.4.2 or later.
pip3 install --upgrade transformers optimum
# If using PyTorch 2.1 + CUDA 12.x:
pip3 install --upgrade auto-gptq
# or, if using PyTorch 2.1 + CUDA 11.x:
pip3 install --upgrade auto-gptq --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/
If you are using PyTorch 2.0, you will need to install AutoGPTQ from source. Likewise if you have problems with the pre-built wheels, you should try building from source:
pip3 uninstall -y auto-gptq
git clone https://github.com/PanQiWei/AutoGPTQ
cd AutoGPTQ
git checkout v0.5.1
pip3 install .
Example Python code
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
model_name_or_path = "TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ"
# To use a different branch, change revision
# For example: revision="gptq-4bit-128g-actorder_True"
model = AutoModelForCausalLM.from_pretrained(model_name_or_path,
device_map="auto",
trust_remote_code=True,
revision="main")
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
prompt = "Tell me about AI"
prompt_template=f'''<|im_start|>system
{system_message}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
'''
print("\n\n*** Generate:")
input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda()
output = model.generate(inputs=input_ids, temperature=0.7, do_sample=True, top_p=0.95, top_k=40, max_new_tokens=512)
print(tokenizer.decode(output[0]))
# Inference can also be done using transformers' pipeline
print("*** Pipeline:")
pipe = pipeline(
"text-generation",
model=model,
tokenizer=tokenizer,
max_new_tokens=512,
do_sample=True,
temperature=0.7,
top_p=0.95,
top_k=40,
repetition_penalty=1.1
)
print(pipe(prompt_template)[0]['generated_text'])
Compatibility
These GPTQs are not yet working.
Discord
For further support, and discussions on these models and AI in general, join us at:
Thanks, and how to contribute
Thanks to Clay from gpus.llm-utils.org!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
- Patreon: https://patreon.com/TheBlokeAI
- Ko-Fi: https://ko-fi.com/TheBlokeAI
Special thanks to: Aemon Algiz.
Patreon special mentions: Michael Levine, 阿明, Trailburnt, Nikolai Manek, John Detwiler, Randy H, Will Dee, Sebastain Graf, NimbleBox.ai, Eugene Pentland, Emad Mostaque, Ai Maven, Jim Angel, Jeff Scroggin, Michael Davis, Manuel Alberto Morcote, Stephen Murray, Robert, Justin Joy, Luke @flexchar, Brandon Frisco, Elijah Stavena, S_X, Dan Guido, Undi ., Komninos Chatzipapas, Shadi, theTransient, Lone Striker, Raven Klaugh, jjj, Cap'n Zoog, Michel-Marie MAUDET (LINAGORA), Matthew Berman, David, Fen Risland, Omer Bin Jawed, Luke Pendergrass, Kalila, OG, Erik Bjäreholt, Rooh Singh, Joseph William Delisle, Dan Lewis, TL, John Villwock, AzureBlack, Brad, Pedro Madruga, Caitlyn Gatomon, K, jinyuan sun, Mano Prime, Alex, Jeffrey Morgan, Alicia Loh, Illia Dulskyi, Chadd, transmissions 11, fincy, Rainer Wilmers, ReadyPlayerEmma, knownsqashed, Mandus, biorpg, Deo Leter, Brandon Phillips, SuperWojo, Sean Connelly, Iucharbius, Jack West, Harry Royden McLaughlin, Nicholas, terasurfer, Vitor Caleffi, Duane Dunston, Johann-Peter Hartmann, David Ziegler, Olakabola, Ken Nordquist, Trenton Dambrowitz, Tom X Nguyen, Vadim, Ajan Kanaga, Leonard Tan, Clay Pascal, Alexandros Triantafyllidis, JM33133, Xule, vamX, ya boyyy, subjectnull, Talal Aujan, Alps Aficionado, wassieverse, Ari Malik, James Bentley, Woland, Spencer Kim, Michael Dempsey, Fred von Graf, Elle, zynix, William Richards, Stanislav Ovsiannikov, Edmond Seymore, Jonathan Leane, Martin Kemka, usrbinkat, Enrico Ros
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
Original model card: Disco Research's Discolm Mixtral 8X7B v2
Eight french experts sitting at a table. There's lots of wind.
DiscoLM Mixtral 8x7b alpha
DiscoLM Mixtral 8x7b alpha is an experimental 8x7b MoE model based on Mistral AI´s Mixtral 8x7b. This model is based on experimental code converting the model weights to huggingface format and enabling Transformers-based inference. It was then finetuned on the Synthia, MethaMathQA und Capybara datasets. DiscoLM Mixtral 8x7b alpha is a DiscoResearch project and was created by Björn Plüster with lots of support from the community.
Many thanks to HessianAI for providing the compute resources for this project and to the great people at LAION without whom this project would not have been possible!
Table of Contents
Download
Please note that you have to run the model with trust_remote_code=True until the new arch is merged into transformers!
| Huggingface | GPTQ | GGUF | AWQ | Base Model |
|---|---|---|---|---|
| Link | tbc | tbc | tbc | tbc |
Benchmarks
Huggingface Leaderboard
This model is still an early Alpha with experimental code and we can't guarantee that there all values are correct. The following are the scores from our own evaluation.
| Metric | Value |
|---|---|
| ARC (25-shot) | 67.32 |
| HellaSwag (10-shot) | 86.25 |
| MMLU (5-shot) | 70.72 |
| TruthfulQA (0-shot) | 54.17 |
| Winogrande (5-shot) | 80.72 |
| GSM8k (5-shot) | 25.09 (bad score. no clue why) |
| Avg. | 64.05 |
We use Language Model Evaluation Harness to run the benchmark tests above, using the same version as the HuggingFace LLM Leaderboard.
FastEval
tbc
MTBench
tbc
Prompt Format
Please note that you have to run the model with trust_remote_code=True until the new arch is merged into transformers!
This model follows the ChatML format:
<|im_start|>system
You are DiscoLM, a helpful assistant.
<|im_end|>
<|im_start|>user
Please tell me possible reasons to call a research collective "Disco Research"<|im_end|>
<|im_start|>assistant
This formatting is also available via a pre-defined Transformers chat template, which means that lists of messages can be formatted for you with the apply_chat_template() method:
chat = [
{"role": "system", "content": "You are DiscoLM, a helpful assistant."},
{"role": "user", "content": "Please tell me possible reasons to call a research collective Disco Research"}
]
tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
If you use tokenize=True and return_tensors="pt" instead, then you will get a tokenized and formatted conversation ready to pass to model.generate().
Basic inference code:
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("DiscoResearch/DiscoLM-mixtral-8x7b-v2", low_cpu_mem_usage=True, device_map="auto", trust_remote_code=True)
tok = AutoTokenizer.from_pretrained("DiscoResearch/DiscoLM-mixtral-8x7b-v2")
chat = [
{"role": "system", "content": "You are DiscoLM, a helpful assistant."},
{"role": "user", "content": "Please tell me possible reasons to call a research collective Disco Research"}
]
x = tokenizer.apply_chat_template(chat, tokenize=True, return_tensors="pt", add_generation_prompt=True).cuda()
x = model.generate(x, max_new_tokens=128).cpu()
print(tok.batch_decode(x))
Datasets
The following datasets were used for training DiscoLM Mixtral 8x7b alpha:
- Synthia
- MetaMathQA
- NousReseach Capybara (currently not public)
Many thanks for all dataset providers/curators!
Contact
Best way to reach us is on our Discord.
About DiscoResearch
DiscoResearch is an aspiring open research community. Disco should be a place where researchers from many communities can come together to combine their expertise and create innovative and groundbreaking LLMs. Come join our Discord, share your opinions and ideas, and advance open LLM research with us!
Acknowledgements
Many thanks first and foremost to Mistral AI for releasing another awesome model and their release strategy that is much fun for the whole community. Additionally, many thanks in particular to Dmytro Dzhulgakov who was the first one with a running inference implementation, Vik who spotted a critical bug in our first implementation (he actually read the paper!), winglian for helpful advice and Axolotl which was used to finetune the model, MigTissera, MetaMath and NousResearch for their great datasets, and everyone who participated in this awesome speedrun on either our, the Nous Research or one of the other Discords (please contact us if we forgot to mention you here!).
DiscoLM Mixtral is a DiscoResearch project and was created by Björn Plüster. The model was trained with compute provided by HessianAI; many thanks as well to LAION for their coordination and providing invaluable contacts + advice.
Disclaimer
The license on this model does not constitute legal advice. We are not responsible for the actions of third parties who use this model. This model should only be used for research purposes.
- Downloads last month
- 48
Model tree for TheBloke/DiscoLM-mixtral-8x7b-v2-GPTQ
Base model
DiscoResearch/DiscoLM-mixtral-8x7b-v2