README.md

metadata

datasets:
  - guanaco
inference: false
language:
  - en
license:
  - apache-2.0
model_hub_library:
  - transformers
model_type: starcoder

Chat & support: my new Discord server

Want to contribute? TheBloke's Patreon page

LoupGarou's WizardCoder-Guanaco-15B-V1.1 GPTQ

These files are GPTQ model files for LoupGarou's WizardCoder-Guanaco-15B-V1.1.

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 models were quantised using hardware kindly provided by Latitude.sh.

Repositories available

• GPTQ models for GPU inference, with multiple quantisation parameter options.
• 4, 5, and 8-bit GGML models for CPU+GPU inference
• Unquantised fp16 model in pytorch format, for GPU inference and for further conversions

Prompt template: Alpaca

Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction: {prompt}

### Response:

Provided files

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.

Branch	Bits	Group Size	Act Order (desc_act)	File Size	ExLlama Compatible?	Made With	Description
main	4	128	False	9.20 GB	False	AutoGPTQ	Most compatible option. Good inference speed in AutoGPTQ and GPTQ-for-LLaMa. Lower inference quality than other options.
gptq-4bit-32g-actorder_True	4	32	True	10.09 GB	False	AutoGPTQ	4-bit, with Act Order and group size. 32g gives highest possible inference quality, with maximum VRAM usage. Poor AutoGPTQ CUDA speed.
gptq-4bit-64g-actorder_True	4	64	True	9.49 GB	False	AutoGPTQ	4-bit, with Act Order and group size. 64g uses less VRAM than 32g, but with slightly lower accuracy. Poor AutoGPTQ CUDA speed.
gptq-4bit-128g-actorder_True	4	128	True	9.20 GB	False	AutoGPTQ	4-bit, with Act Order and group size. 128g uses even less VRAM, but with slightly lower accuracy. Poor AutoGPTQ CUDA speed.
gptq-8bit--1g-actorder_True	8	None	True	16.49 GB	False	AutoGPTQ	8-bit, with Act Order. No group size, to lower VRAM requirements and to improve AutoGPTQ speed.
gptq-8bit-128g-actorder_True	8	128	True	16.84 GB	False	AutoGPTQ	8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. Poor AutoGPTQ CUDA speed.
gptq-8bit-128g-actorder_False	8	128	False	16.84 GB	False	AutoGPTQ	8-bit, with group size 128g for higher inference quality and without Act Order to improve AutoGPTQ speed.

How to download from branches

• In text-generation-webui, you can add :branch to the end of the download name, eg TheBloke/WizardCoder-Guanaco-15B-V1.1-GPTQ:gptq-4bit-32g-actorder_True
• With Git, you can clone a branch with:

git clone --branch gptq-4bit-32g-actorder_True https://huggingface.co/TheBloke/WizardCoder-Guanaco-15B-V1.1-GPTQ`

• In Python Transformers code, the branch is the revision parameter; see below.

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 know how to make a manual install.

1. Click the Model tab.
2. Under Download custom model or LoRA, enter TheBloke/WizardCoder-Guanaco-15B-V1.1-GPTQ.

• To download from a specific branch, enter for example TheBloke/WizardCoder-Guanaco-15B-V1.1-GPTQ:gptq-4bit-32g-actorder_True
• see Provided Files above for the list of branches for each option.

3. Click Download.
4. The model will start downloading. Once it's finished it will say "Done"
5. In the top left, click the refresh icon next to Model.
6. In the Model dropdown, choose the model you just downloaded: WizardCoder-Guanaco-15B-V1.1-GPTQ
7. The model will automatically load, and is now ready for use!
8. 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 set GPTQ parameters any more. These are set automatically from the file quantize_config.json.

9. 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

First make sure you have AutoGPTQ installed:

GITHUB_ACTIONS=true pip install auto-gptq

Then try the following example code:

from transformers import AutoTokenizer, pipeline, logging
from auto_gptq import AutoGPTQForCausalLM, BaseQuantizeConfig

model_name_or_path = "TheBloke/WizardCoder-Guanaco-15B-V1.1-GPTQ"
model_basename = "gptq_model-4bit-128g"

use_triton = False

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

model = AutoGPTQForCausalLM.from_quantized(model_name_or_path,
        model_basename=model_basename
        use_safetensors=True,
        trust_remote_code=False,
        device="cuda:0",
        use_triton=use_triton,
        quantize_config=None)

"""
To download from a specific branch, use the revision parameter, as in this example:

model = AutoGPTQForCausalLM.from_quantized(model_name_or_path,
        revision="gptq-4bit-32g-actorder_True",
        model_basename=model_basename,
        use_safetensors=True,
        trust_remote_code=False,
        device="cuda:0",
        quantize_config=None)
"""

prompt = "Tell me about AI"
prompt_template=f'''Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction: {prompt}

### Response:
'''

print("\n\n*** Generate:")

input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda()
output = model.generate(inputs=input_ids, temperature=0.7, max_new_tokens=512)
print(tokenizer.decode(output[0]))

# Inference can also be done using transformers' pipeline

# Prevent printing spurious transformers error when using pipeline with AutoGPTQ
logging.set_verbosity(logging.CRITICAL)

print("*** Pipeline:")
pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    max_new_tokens=512,
    temperature=0.7,
    top_p=0.95,
    repetition_penalty=1.15
)

print(pipe(prompt_template)[0]['generated_text'])

Compatibility

The files provided will work with AutoGPTQ (CUDA and Triton modes), GPTQ-for-LLaMa (only CUDA has been tested), and Occ4m's GPTQ-for-LLaMa fork.

ExLlama works with Llama models in 4-bit. Please see the Provided Files table above for per-file compatibility.

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!

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: Luke from CarbonQuill, Aemon Algiz.

Patreon special mentions: Space Cruiser, Nikolai Manek, Sam, Chris McCloskey, Rishabh Srivastava, Kalila, Spiking Neurons AB, Khalefa Al-Ahmad, WelcomeToTheClub, Chadd, Lone Striker, Viktor Bowallius, Edmond Seymore, Ai Maven, Chris Smitley, Dave, Alexandros Triantafyllidis, Luke @flexchar, Elle, ya boyyy, Talal Aujan, Alex , Jonathan Leane, Deep Realms, Randy H, subjectnull, Preetika Verma, Joseph William Delisle, Michael Levine, chris gileta, K, Oscar Rangel, LangChain4j, Trenton Dambrowitz, Eugene Pentland, Johann-Peter Hartmann, Femi Adebogun, Illia Dulskyi, senxiiz, Daniel P. Andersen, Sean Connelly, Artur Olbinski, RoA, Mano Prime, Derek Yates, Raven Klaugh, David Flickinger, Willem Michiel, Pieter, Willian Hasse, vamX, Luke Pendergrass, webtim, Ghost , Rainer Wilmers, Nathan LeClaire, Will Dee, Cory Kujawski, John Detwiler, Fred von Graf, biorpg, Iucharbius , Imad Khwaja, Pierre Kircher, terasurfer , Asp the Wyvern, John Villwock, theTransient, zynix , Gabriel Tamborski, Fen Risland, Gabriel Puliatti, Matthew Berman, Pyrater, SuperWojo, Stephen Murray, Karl Bernard, Ajan Kanaga, Greatston Gnanesh, Junyu Yang.

Thank you to all my generous patrons and donaters!

Original model card: LoupGarou's WizardCoder-Guanaco-15B-V1.1

WizardCoder-Guanaco-15B-V1.1 Model Card

The WizardCoder-Guanaco-15B-V1.1 is a language model that combines the strengths of the WizardCoder base model and the openassistant-guanaco dataset for finetuning. The openassistant-guanaco dataset was further trimmed to within 2 standard deviations of token size for input and output pairs and all non-english data has been removed to reduce training size requirements.

Version 1.1 showcases notable enhancements, employing a modified version of the previous openassistant-guanaco dataset. This dataset underwent a comprehensive revision, replacing every single answer with those generated by GPT-4.

The volume of the datasets has also been augmented by approximately 50%, with a particular focus on high school and abstract algebra. This expansion leveraged the combined capabilities of GPT-4 and GPT-3.5-Turbo. The initial evaluation of algebraic functions over 12 epochs indicated promising results from this enriched dataset. However, this is just the beginning; further refinements are in the pipeline, aiming to optimize the dataset quality and subsequently decrease the number of epochs required to achieve comparable results.

Considering the need to curtail memory consumption during training, this dataset was tailored to consist solely of English language questions and answers. Consequently, the model's performance in language translation may not be up to par. Nevertheless, the focus remains on enhancing the model's proficiency and efficiency within its defined scope.

Intended Use

This model is designed to be used for a wide array of text generation tasks that require understanding and generating English text. The model is expected to perform well in tasks such as answering questions, writing essays, summarizing text, translation, and more. However, given the specific data processing and finetuning done, it might be particularly effective for tasks related to English language question-answering systems.

Limitations

Despite the powerful capabilities of this model, users should be aware of its limitations. The model's knowledge is up to date only until the time it was trained, and it doesn't know about events in the world after that. It can sometimes produce incorrect or nonsensical responses, as it doesn't understand the text in the same way humans do. It should be used as a tool to assist in generating text and not as a sole source of truth.

How to use

Here is an example of how to use this model:

from transformers import AutoModelForCausalLM, AutoTokenizer
import time
import torch

class Chatbot:
    def __init__(self, model_name):
        self.tokenizer = AutoTokenizer.from_pretrained(model_name, padding_side='left')
        self.model = AutoModelForCausalLM.from_pretrained(model_name, load_in_4bit=True, torch_dtype=torch.bfloat16)
        if self.tokenizer.pad_token_id is None:
            self.tokenizer.pad_token_id = self.tokenizer.eos_token_id

    def get_response(self, prompt):
        inputs = self.tokenizer.encode_plus(prompt, return_tensors="pt", padding='max_length', max_length=100)
        if next(self.model.parameters()).is_cuda:
            inputs = {name: tensor.to('cuda') for name, tensor in inputs.items()}
        start_time = time.time()
        tokens = self.model.generate(input_ids=inputs['input_ids'], 
                                    attention_mask=inputs['attention_mask'],
                                    pad_token_id=self.tokenizer.pad_token_id,
                                    max_new_tokens=400)
        end_time = time.time()
        output_tokens = tokens[0][inputs['input_ids'].shape[-1]:]
        output = self.tokenizer.decode(output_tokens, skip_special_tokens=True)
        time_taken = end_time - start_time
        return output, time_taken

def main():
    chatbot = Chatbot("LoupGarou/WizardCoder-Guanaco-15B-V1.1")
    while True:
        user_input = input("Enter your prompt: ")
        if user_input.lower() == 'quit':
            break
        output, time_taken = chatbot.get_response(user_input)
        print("\033[33m" + output + "\033[0m")
        print("Time taken to process: ", time_taken, "seconds")
    print("Exited the program.")

if __name__ == "__main__":
    main()

Training Procedure

The WizardCoder model, serving as the base, was fine-tuned on a modified version of the openassistant-guanaco dataset. This dataset underwent a significant revision, replacing every single answer with responses generated by the AI model GPT-4. It was then expanded by approximately 50%, emphasizing high school and abstract algebra-related questions, using a mix of GPT-4 and GPT-3.5-Turbo for answer generation.

The selected dataset was standardized to fall within two standard deviations of token size for the question sets, ensuring consistency in data handling. The order of the questions was also randomized to mitigate any potential biases during the training phase.

In the interest of optimizing memory usage during the training process, the dataset was streamlined to only include English language content. As a result, all non-English data was systematically expunged from this fine-tuning dataset. It's worth noting that this modification limits the model's performance in language translation tasks, but it significantly boosts its efficiency and effectiveness when dealing with English language questions and answers.

Acknowledgements

This model, WizardCoder-Guanaco-15B-V1.1, is simply building on the efforts of two great teams to evaluate the performance of a combined model with the strengths of the WizardCoder base model and the openassistant-guanaco dataset.

A sincere appreciation goes out to the developers and the community involved in the creation and refinement of these models. Their commitment to providing open source tools and datasets have been instrumental in making this project a reality.

Moreover, a special note of thanks to the Hugging Face team, whose transformative library has not only streamlined the process of model creation and adaptation, but also democratized the access to state-of-the-art machine learning technologies. Their impact on the development of this project cannot be overstated.