Want to contribute? TheBloke's Patreon page

MosaicML's MPT-30B-Instruct GGML

These files are GGML format model files for MosaicML's MPT-30B-Instruct.

Please note that these GGMLs are not compatible with llama.cpp, or currently with text-generation-webui. Please see below for a list of tools known to work with these model files.

KoboldCpp just added GPU accelerated (OpenCL) support for MPT models, so that is the client I recommend using for these models.

Note: Please use version 1.32.1 or later of KoboldCpp, as there was a bug in 1.32 that affected loading of MPT models.

Repositories available

Prompt template

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

### Instruction: prompt

### Response:

A note regarding context length: 8K

The base model has an 8K context length. It is not yet confirmed if the 8K context of this model works with the quantised files.

If it does, KoboldCpp supports 8K context if you manually set it to 8K by adjusting the text box above the slider:

It is currently unknown as to increased context is compatible with other MPT GGML clients.

If you have feedback on this, please let me know.

Compatibilty

These files are not compatible with text-generation-webui, llama.cpp, or llama-cpp-python.

Currently they can be used with:

KoboldCpp, a powerful inference engine based on llama.cpp, with good UI and GPU accelerated support for MPT models: KoboldCpp
The ctransformers Python library, which includes LangChain support: ctransformers
The LoLLMS Web UI which uses ctransformers: LoLLMS Web UI
rustformers' llm
The example mpt binary provided with ggml

As other options become available I will endeavour to update them here (do let me know in the Community tab if I've missed something!)

Tutorial for using LoLLMS Web UI

Provided files

Name	Quant method	Bits	Size	Max RAM required	Use case
mpt-30b-instruct.ggmlv0.q4_0.bin	q4_0	4	16.85 GB	19.35 GB	Original llama.cpp quant method, 4-bit.
mpt-30b-instruct.ggmlv0.q4_1.bin	q4_1	4	18.73 GB	21.23 GB	Original llama.cpp quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models.
mpt-30b-instruct.ggmlv0.q5_0.bin	q5_0	5	20.60 GB	23.10 GB	Original llama.cpp quant method, 5-bit. Higher accuracy, higher resource usage and slower inference.
mpt-30b-instruct.ggmlv0.q5_1.bin	q5_1	5	22.47 GB	24.97 GB	Original llama.cpp quant method, 5-bit. Even higher accuracy, resource usage and slower inference.
mpt-30b-instruct.ggmlv0.q8_0.bin	q8_0	8	31.83 GB	34.33 GB	Original llama.cpp quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users.

Note: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.

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, Dmitriy Samsonov.

Patreon special mentions: Mano Prime, Fen Risland, Derek Yates, Preetika Verma, webtim, Sean Connelly, Alps Aficionado, Karl Bernard, Junyu Yang, Nathan LeClaire, Chris McCloskey, Lone Striker, Asp the Wyvern, Eugene Pentland, Imad Khwaja, trip7s trip, WelcomeToTheClub, John Detwiler, Artur Olbinski, Khalefa Al-Ahmad, Trenton Dambrowitz, Talal Aujan, Kevin Schuppel, Luke Pendergrass, Pyrater, Joseph William Delisle, terasurfer , vamX, Gabriel Puliatti, David Flickinger, Jonathan Leane, Iucharbius , Luke, Deep Realms, Cory Kujawski, ya boyyy, Illia Dulskyi, senxiiz, Johann-Peter Hartmann, John Villwock, K, Ghost , Spiking Neurons AB, Nikolai Manek, Rainer Wilmers, Pierre Kircher, biorpg, Space Cruiser, Ai Maven, subjectnull, Willem Michiel, Ajan Kanaga, Kalila, chris gileta, Oscar Rangel.

Thank you to all my generous patrons and donaters!

Original model card: MosaicML's MPT-30B-Instruct

MPT-30B-Instruct

MPT-30B-Instruct is a model for short-form instruction following. It is built by finetuning MPT-30B on Dolly HHRLHF derived from the Databricks Dolly-15k and the Anthropic Helpful and Harmless (HH-RLHF) datasets. It is also trained on Competition Math, Duorc, CoT GSM8k, Qasper, Quality, Summ Screen FD and Spider.

License: CC-By-SA-3.0

This model was trained by MosaicML and follows a modified decoder-only transformer architecture.

Model Date

June 22, 2023

Model License

CC-By-SA-3.0

Documentation

Blog post: Raising the bar for open-source foundation models
Codebase (mosaicml/llm-foundry repo)
Questions: Feel free to contact us via the MosaicML Community Slack!

Example Question/Instruction

Bespokenizer46

I need to convince my friend, Phyllis, that she should train a custom LLM for her Fortune 500 company using the MosaicML Platform. Please write an email that explains why MosaicML's emphasis on cutting edge methodology, data privacy, and efficiency are so important. End the email with a friendly inquiry about Phyllis's family.

MPT-30B-Instruct:

Phyllis - I know you're always looking for ways to improve efficiency at work and I think I've found a solution that you'll be very interested in. LLMs are only as good as the data they're trained on and the training process can be very time consuming and resource intensive. MosaicML uses cutting edge methodology to efficiently train high quality models without compromising data privacy. They also provide tools to easily connect to and use the model in your daily workflow. I think you'd really enjoy speaking with their founder, we can set up a call if you're interested. Also, I know it's been a tough year for your family, how are things?

Best, Your Friend

How to Use

Note: This model requires that trust_remote_code=True be passed to the from_pretrained method. This is because we use a custom model architecture that is not yet part of the transformers package.

It includes options for many training efficiency features such as FlashAttention (Dao et al. 2022), ALiBi, QK LayerNorm, and more.

import transformers
model = transformers.AutoModelForCausalLM.from_pretrained(
  'mosaicml/mpt-30b-instruct',
  trust_remote_code=True
)

To use the optimized triton implementation of FlashAttention, you can load the model on GPU (cuda:0) with attn_impl='triton' and with bfloat16 precision:

import torch
import transformers

name = 'mosaicml/mpt-30b-instruct'

config = transformers.AutoConfig.from_pretrained(name, trust_remote_code=True)
config.attn_config['attn_impl'] = 'triton'  # change this to use triton-based FlashAttention
config.init_device = 'cuda:0' # For fast initialization directly on GPU!

model = transformers.AutoModelForCausalLM.from_pretrained(
  name,
  config=config,
  torch_dtype=torch.bfloat16, # Load model weights in bfloat16
  trust_remote_code=True
)

The model was trained initially on a sequence length of 2048. An additional pre-training phase was included for sequence length adaptation to 8192. However, ALiBi further enables users to increase the maximum sequence length during finetuning and/or inference. For example:

import transformers

name = 'mosaicml/mpt-30b-instruct'

config = transformers.AutoConfig.from_pretrained(name, trust_remote_code=True)
config.max_seq_len = 16384 # (input + output) tokens can now be up to 16384

model = transformers.AutoModelForCausalLM.from_pretrained(
  name,
  config=config,
  trust_remote_code=True
)

This model was trained with the MPT-30B tokenizer which is based on the EleutherAI/gpt-neox-20b tokenizer and includes additional padding and eos tokens.

from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained('mosaicml/mpt-30b')

The model can then be used, for example, within a text-generation pipeline.
Note: when running Torch modules in lower precision, it is best practice to use the torch.autocast context manager.

from transformers import pipeline

with torch.autocast('cuda', dtype=torch.bfloat16):
    inputs = tokenizer('Here is a recipe for vegan banana bread:\n', return_tensors="pt").to('cuda')
    outputs = model.generate(**inputs, max_new_tokens=100)
    print(tokenizer.batch_decode(outputs, skip_special_tokens=True))

# or using the HF pipeline
pipe = pipeline('text-generation', model=model, tokenizer=tokenizer, device='cuda:0')
with torch.autocast('cuda', dtype=torch.bfloat16):
    print(
        pipe('Here is a recipe for vegan banana bread:\n',
            max_new_tokens=100,
            do_sample=True,
            use_cache=True))

Formatting

This model was trained on data formatted as follows:

def format_prompt(instruction):
    template = "Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n###Instruction\n{instruction}\n\n### Response\n"
    return template.format(instruction=instruction)
)

example = "Tell me a funny joke.\nDon't make it too funny though."
fmt_ex = format_prompt(instruction=example)

In the above example, fmt_ex is ready to be tokenized and sent through the model.

Model Description

The architecture is a modification of a standard decoder-only transformer.

The model has been modified from a standard transformer in the following ways:

It uses FlashAttention
It uses ALiBi (Attention with Linear Biases) and does not use positional embeddings
It does not use biases

Hyperparameter	Value
n_parameters	29.95B
n_layers	48
n_heads	64
d_model	7168
vocab size	50432
sequence length	8192

Data Mix

The model was trained on the following data mix:

Data Source	Number of Tokens in Source	Proportion
competition_math	1.6 M	3.01%
cot_gsm8k	3.36 M	6.32%
dialogsum	0.1 M	0.19%
dolly_hhrlhf	5.89 M	11.07%
duorc	8.2 M	15.51%
qasper	10.97 M	20.63%
quality	11.31 M	21.28%
scrolls/summ_screen_fd	11.56 M	21.82%
spider	0.089 M	0.16%

PreTraining Data

For more details on the pretraining process, see MPT-30B.

The data was tokenized using the EleutherAI/gpt-neox-20b tokenizer.

Training Configuration

This model was trained on 72 A100 40GB GPUs for 8 hours using the MosaicML Platform. The model was trained with sharded data parallelism using FSDP and used the AdamW optimizer.

Limitations and Biases

The following language is modified from EleutherAI's GPT-NeoX-20B

MPT-30B-Instruct can produce factually incorrect output, and should not be relied on to produce factually accurate information. MPT-30B-Instruct was trained on various public datasets. While great efforts have been taken to clean the pretraining data, it is possible that this model could generate lewd, biased or otherwise offensive outputs.

Acknowledgements

This model was finetuned by Sam Havens, Alex Trott, and the MosaicML NLP team

MosaicML Platform

If you're interested in training and deploying your own MPT or LLMs on the MosaicML Platform, sign up here.

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. Please consult an attorney before using this model for commercial purposes.

Citation

Please cite this model using the following format:

@online{MosaicML2023Introducing,
    author    = {MosaicML NLP Team},
    title     = {Introducing MPT-30B: Raising the bar
for open-source foundation models},
    year      = {2023},
    url       = {www.mosaicml.com/blog/mpt-30b},
    note      = {Accessed: 2023-06-22},
    urldate   = {2023-06-22}
}