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.
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.
All recent GPTQ files are made with AutoGPTQ, and all files in non-main branches are made with AutoGPTQ. Files in the main branch which were uploaded before August 2023 were made with GPTQ-for-LLaMa.
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 dataset used for quantisation. Using a dataset more appropriate to the model's training can improve quantisation accuracy. Note that the GPTQ 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 models in 4-bit.
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/Airoboros-L2-13B-2_1-YaRN-64K-GPTQ.
To download from a specific branch, enter for example TheBloke/Airoboros-L2-13B-2_1-YaRN-64K-GPTQ:main
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: Airoboros-L2-13B-2_1-YaRN-64K-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!
How to use this GPTQ model from Python code
Install the necessary packages
Requires: Transformers 4.32.0 or later, Optimum 1.12.0 or later, and AutoGPTQ 0.4.2 or later.
shell
1pip3 install transformers>=4.32.0 optimum>=1.12.0
2pip3 install auto-gptq --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/ # Use cu117 if on CUDA 11.7
If you have problems installing AutoGPTQ using the pre-built wheels, install it from source instead:
1from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
23model_name_or_path ="TheBloke/Airoboros-L2-13B-2_1-YaRN-64K-GPTQ"4# To use a different branch, change revision5# For example: revision="main"6model = AutoModelForCausalLM.from_pretrained(model_name_or_path,7 device_map="auto",8 trust_remote_code=True,9 revision="main")1011tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)1213prompt ="Tell me about AI"14prompt_template=f'''A chat.
15USER: {prompt}16ASSISTANT:
1718'''1920print("\n\n*** Generate:")2122input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda()23output = model.generate(inputs=input_ids, temperature=0.7, do_sample=True, top_p=0.95, top_k=40, max_new_tokens=512)24print(tokenizer.decode(output[0]))2526# Inference can also be done using transformers' pipeline2728print("*** Pipeline:")29pipe = pipeline(30"text-generation",31 model=model,32 tokenizer=tokenizer,33 max_new_tokens=512,34 do_sample=True,35 temperature=0.7,36 top_p=0.95,37 top_k=40,38 repetition_penalty=1.139)4041print(pipe(prompt_template)[0]['generated_text'])
Compatibility
The files provided are tested to work with AutoGPTQ, both via Transformers and using AutoGPTQ directly. They should also work with Occ4m's GPTQ-for-LLaMa fork.
ExLlama is compatible with Llama models in 4-bit. Please see the Provided Files table above for per-file compatibility.
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 special mentions: Alicia Loh, Stephen Murray, K, Ajan Kanaga, RoA, Magnesian, Deo Leter, Olakabola, Eugene Pentland, zynix, Deep Realms, Raymond Fosdick, Elijah Stavena, Iucharbius, Erik Bjäreholt, Luis Javier Navarrete Lozano, Nicholas, theTransient, John Detwiler, alfie_i, knownsqashed, Mano Prime, Willem Michiel, Enrico Ros, LangChain4j, OG, Michael Dempsey, Pierre Kircher, Pedro Madruga, James Bentley, Thomas Belote, Luke @flexchar, Leonard Tan, Johann-Peter Hartmann, Illia Dulskyi, Fen Risland, Chadd, S_X, Jeff Scroggin, Ken Nordquist, Sean Connelly, Artur Olbinski, Swaroop Kallakuri, Jack West, Ai Maven, David Ziegler, Russ Johnson, transmissions 11, John Villwock, Alps Aficionado, Clay Pascal, Viktor Bowallius, Subspace Studios, Rainer Wilmers, Trenton Dambrowitz, vamX, Michael Levine, 준교 김, Brandon Frisco, Kalila, Trailburnt, Randy H, Talal Aujan, Nathan Dryer, Vadim, 阿明, ReadyPlayerEmma, Tiffany J. Kim, George Stoitzev, Spencer Kim, Jerry Meng, Gabriel Tamborski, Cory Kujawski, Jeffrey Morgan, Spiking Neurons AB, Edmond Seymore, Alexandros Triantafyllidis, Lone Striker, Cap'n Zoog, Nikolai Manek, danny, ya boyyy, Derek Yates, usrbinkat, Mandus, TL, Nathan LeClaire, subjectnull, Imad Khwaja, webtim, Raven Klaugh, Asp the Wyvern, Gabriel Puliatti, Caitlyn Gatomon, Joseph William Delisle, Jonathan Leane, Luke Pendergrass, SuperWojo, Sebastain Graf, Will Dee, Fred von Graf, Andrey, Dan Guido, Daniel P. Andersen, Nitin Borwankar, Elle, Vitor Caleffi, biorpg, jjj, NimbleBox.ai, Pieter, Matthew Berman, terasurfer, Michael Davis, Alex, Stanislav Ovsiannikov
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
Original model card: bhenrym14's Airoboros L2 13B 2.1 YaRN 64K
Extended Context (via YaRN) Finetune of Llama-2-13b with airoboros-2.1 (fp16)
This is a finetune of NousResearch/Yarn-Llama-2-13b-64k, which is base Llama-2-13b with additional pretraining done with YaRN scaling applied to RoPE to extend the useful context length to 64k tokens. Starting with this model, I performed instruction tuning with Jon Durbin's Airoboros 2.1 dataset, with the same scaling approach applied.
This is a (merged) QLoRA fine-tune (rank 64).
The finetune was performed with 1x RTX 6000 Ada (~16 hours).
How to Use
YaRN is not implemented natively in Transformers. The YaRN pretrained model NousResearch/Yarn-Llama-2-13b-64k contains a drop-in llama architecture replacement that interfaces with the included configuration file. To maximize compatibility, I have included the version that omits flash attention. To run using Transformers, you will therefore need to pass trust_remote_code=True.
The PNTK method employed in my other model bhenrym14/airophin-13b-pntk-16k-fp16, is very similar to YaRN. For GPTQ, I have an exllama patch that I may adapt for YaRN, but the community appears motivated to rapidly implement YaRN in common libraries, so I may not bother.
Please comment with any questions and feedback on how this model performs, especially at long context lengths!
Ooba use: Be sure to increase the Truncate the prompt up to this length parameter to 65586 to utilize the full context capabilities. Again trust_remote_code=True is imperative. Obviously, using full context requires A LOT of VRAM.
There may be issues on Windows systems loading this model due to the decimal in "2.1" found in the model name. Try simply changing the model directory name to omit this decimal if you have issues loading the model.
Motivation
Yet another RoPE extensioN method (YaRN) is a novel method of extending the useful context of pretrained LLMs, with architectures employing RoPE, with minimal additonal training requirements. This method is the consequence of efforts to mitigate the shortcomings of other methods such as Position Interpolation (PI) and NTK-Aware scaling. This model is an attempt to enable the community to assess the capabilities of this extension method in real world applications.
Despite having a far higher scaling factor, this model is competitive with bhenrym14/airophin-13b-pntk-16k-fp16 at short context lengths.
I may need to restrict these comparisons to models finetuned on the same dataset. Differences between airoboros 1.4.1 and 2.0m/2.1 may be a confounder.
Overall, it appears that YaRN is capable of extending the context window with minimal impact to short context performance, when compared to other methods. Furthermore, it's able to do this with a FAR higher scaling factor, which with other methods (especially PI), resulted in serious performance degradation at shorter context lengths.
Both the YaRN and Code LLama papers suggest that YaRN and NTK scaling may ameliorate the issue of "U shaped" attention to some degree, where long context models struggle to attend to information in the middle of the context window. Further study is needed to evaluate this. Anecdotal feedback from the community on this issue would be appreciated!
