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.

These files were quantised using hardware kindly provided by Massed Compute.

Repositories available

Prompt template: Orca-Vicuna

SYSTEM: {system_message}
USER: {prompt}
ASSISTANT:

Known compatible clients / servers

GPTQ models are currently supported on Linux (NVidia/AMD) and Windows (NVidia only). macOS users: please use GGUF models.

These GPTQ models are known to work in the following inference servers/webuis.

This may not be a complete list; if you know of others, please let me know!

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	8192	18.60 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	8192	19.25 GB	Yes	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	8192	21.21 GB	Yes	4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage.
gptq-3bit-128g-actorder_True	3	128	Yes	0.1	VMware Open Instruct	8192	15.03 GB	No	3-bit, with group size 128g and act-order. Higher quality than 128g-False.
gptq-8bit--1g-actorder_True	8	None	Yes	0.1	VMware Open Instruct	8192	35.34 GB	No	8-bit, with Act Order. No group size, to lower VRAM requirements.
gptq-3bit-32g-actorder_True	3	32	Yes	0.1	VMware Open Instruct	8192	16.90 GB	No	3-bit, with group size 64g and act-order. Highest quality 3-bit option.
gptq-8bit-128g-actorder_True	8	128	Yes	0.1	VMware Open Instruct	8192	36.11 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/Yi-34B-200K-DARE-merge-v5-GPTQ in the "Download model" box.

To download from another branch, add :branchname to the end of the download name, eg TheBloke/Yi-34B-200K-DARE-merge-v5-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 Yi-34B-200K-DARE-merge-v5-GPTQ:

shell

mkdir Yi-34B-200K-DARE-merge-v5-GPTQ
huggingface-cli download TheBloke/Yi-34B-200K-DARE-merge-v5-GPTQ --local-dir Yi-34B-200K-DARE-merge-v5-GPTQ --local-dir-use-symlinks False

To download from a different branch, add the --revision parameter:

shell

mkdir Yi-34B-200K-DARE-merge-v5-GPTQ
huggingface-cli download TheBloke/Yi-34B-200K-DARE-merge-v5-GPTQ --revision gptq-4bit-128g-actorder_True --local-dir Yi-34B-200K-DARE-merge-v5-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:

shell

mkdir Yi-34B-200K-DARE-merge-v5-GPTQ
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/Yi-34B-200K-DARE-merge-v5-GPTQ --local-dir Yi-34B-200K-DARE-merge-v5-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/Yi-34B-200K-DARE-merge-v5-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

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/Yi-34B-200K-DARE-merge-v5-GPTQ.
- To download from a specific branch, enter for example TheBloke/Yi-34B-200K-DARE-merge-v5-GPTQ:gptq-4bit-128g-actorder_True
- see Provided Files above for the list of branches for each option.
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: Yi-34B-200K-DARE-merge-v5-GPTQ
The 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.
Once you're ready, click the Text Generation tab and enter a prompt to get started!

Serving this model from Text Generation Inference (TGI)

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/Yi-34B-200K-DARE-merge-v5-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

python

from huggingface_hub import InferenceClient

endpoint_url = "https://your-endpoint-url-here"

prompt = "Tell me about AI"
prompt_template=f'''SYSTEM: {system_message}
USER: {prompt}
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

Install the necessary packages

Requires: Transformers 4.33.0 or later, Optimum 1.12.0 or later, and AutoGPTQ 0.4.2 or later.

shell

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:

shell

pip3 uninstall -y auto-gptq
git clone https://github.com/PanQiWei/AutoGPTQ
cd AutoGPTQ
git checkout v0.5.1
pip3 install .

Example Python code

python

from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

model_name_or_path = "TheBloke/Yi-34B-200K-DARE-merge-v5-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=False,
                                             revision="main")

tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)

prompt = "Write a story about llamas"
system_message = "You are a story writing assistant"
prompt_template=f'''SYSTEM: {system_message}
USER: {prompt}
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

The files provided are tested to work with Transformers. For non-Mistral models, AutoGPTQ can also be used directly.

ExLlama is compatible with Llama architecture models (including Mistral, Yi, DeepSeek, SOLAR, etc) in 4-bit. Please see the Provided Files table above for per-file compatibility.

For a list of clients/servers, please see "Known compatible clients / servers", above.

Discord

For further support, and discussions on these models and AI in general, join us at:

TheBloke AI's Discord server

Thanks, and how to contribute

Thanks to the chirper.ai team!

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: brucethemoose's Yi 34B 200K DARE Merge v5

Nous-Capybara-34B, Tess-M-v1.4, Airoboros-3_1-yi-34b-200k, PlatYi-34B-200K-Q, Pallas-0.4, Yi-34B-200K-AEZAKMI-v2, and a tiny bit of SUS-Chat-34B merged with a new, experimental implementation of "dare ties" via mergekit. See:

Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch

https://github.com/cg123/mergekit/tree/dare

Prompt template: Orca-Vicuna

SYSTEM: {system_message}
USER: {prompt}
ASSISTANT:

It might recognize ChatML, or maybe Llama-chat from Airoboros.

Sometimes the model "spells out" the stop token as </s> like Capybara, so you may need to add </s> as an additional stopping condition.

Running

Being a Yi model, try running a lower temperature with 0.05-0.1 MinP, a little repetition penalty, and no other samplers. Yi tends to run "hot" by default, and it really needs MinP to cull the huge vocabulary.

24GB GPUs can run Yi-34B-200K models at 45K-75K context with exllamav2, and performant UIs like exui. I go into more detail in this post

I recommend exl2 quantizations profiled on data similar to the desired task. It is especially sensitive to the quantization data at low bpw. I've published my own fiction-oriented quantizations here: https://huggingface.co/collections/brucethemoose/most-recent-merge-65742644ca03b6c514afa204

To load this in full-context backends like transformers, you must change max_position_embeddings in config.json to a lower value than 200,000, otherwise you will OOM!

Testing Notes

Merged in mergekit with the following config, and the tokenizer from chargoddard's Yi-Llama:

models:
  - model: /home/alpha/Storage/Models/Raw/chargoddard_Yi-34B-200K-Llama
    # No parameters necessary for base model
  - model: /home/alpha/Storage/Models/Raw/migtissera_Tess-34B-v1.4
    # Less weight than previous merge since Pallas is a finetune of Tess
    parameters:
      weight: 0.14
      density: 0.62
  - model: /home/alpha/FastModels/Mihaiii_Pallas-0.4
    parameters:
      weight: 0.14
      density: 0.62
  - model: /home/alpha//Storage/Models/Raw/bhenrym14_airoboros-3_1-yi-34b-200k
    parameters:
      weight: 0.14
      density: 0.52
  - model: /home/alpha/Storage/Models/Raw/Nous-Capybara-34B
    parameters:
      weight: 0.22
      density: 0.62
  - model: /home/alpha/Storage/Models/Raw/kyujinpy_PlatYi-34B-200k-Q-FastChat
    parameters:
      weight: 0.14
      density: 0.52
  #- model: /home/alpha/Storage/Models/Raw/ehartford_dolphin-2.2-yi-34b-200k
  #  Dolphin 200K seems to be broken according to multiple leaderboards and perplexity tests?
  #  parameters:
  #    weight: 0.15
  #    density: 0.6
  - model: /home/alpha/Models/Raw/adamo1139_Yi-34B-200K-AEZAKMI-v2
    parameters:
      weight: 0.14
      density: 0.52
  - model: /home/alpha/Models/Raw/SUSTech_SUS-Chat-34B/
  # Very low density and low weight since its a Yi 4K finetune, to try and preserve long context performance while "keeping" some of SUS
    parameters:
      weight: 0.08
      density: 0.08
merge_method: dare_ties
base_model: /home/alpha/Storage/Models/Raw/chargoddard_Yi-34B-200K-Llama
parameters:

  int8_mask: true
dtype: bfloat16

Various densities were tested with perplexity tests and long context prompts. Relatively high densities seem to perform better, contrary to the findings of the Super Mario paper.

This particular version is merged with more than the "recommended" max density of 0.5. It seems to result in even better perplexity, but I'm not sure if this translates to better output.

Weights that add up to 1 seems to be optimal.

Dare Ties is also resulting in seemingly better, lower perplexity merges than a regular ties merge, task arithmetic or a slerp merge.

SUS Chat is not a 200K model, hence it was merged at a very low density to try and preserve Yi 200K's long context performance while still inheriting some of SUS's performance.

Dolphin 200K was taken out of this merge because it seems to be performing poorly for a 34B Dolphin model, like something went wrong during training?

I chose not to include other finetunes because they aren't trained on the 200K base. If any other 200K finetunes pop up, let me know.