failspy/mixtral-8x22b-v0.1-instruct-oh-GGUF

GGUF-based Quantized models from fireworks-ai/mixtral-8x22b-instruct-oh

Rest of this readme is a template README.md derived from TheBloke's README

Description

This repo contains GGUF format model files for fireworks-ai/mixtral-8x22b-instruct-oh.

About GGUF

GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.

Here is an incomplete list of clients and libraries that are known to support GGUF:

• llama.cpp. The source project for GGUF. Offers a CLI and a server option.
• text-generation-webui, the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
• KoboldCpp, a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
• GPT4All, a free and open source local running GUI, supporting Windows, Linux and macOS with full GPU accel.
• LM Studio, an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration. Linux available, in beta as of 27/11/2023.
• LoLLMS Web UI, a great web UI with many interesting and unique features, including a full model library for easy model selection.
• Faraday.dev, an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
• llama-cpp-python, a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
• candle, a Rust ML framework with a focus on performance, including GPU support, and ease of use.
• ctransformers, a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server. Note, as of time of writing (November 27th 2023), ctransformers has not been updated in a long time and does not support many recent models.

The model uses Vicuna chat template:

USER: ...

ASSISTANT: ...

or:

SYSTEM: ...

USER: ...

ASSISTANT: ...

Compatibility

These quantised GGUFv2 files are compatible with llama.cpp from March 19th onwards, as of commit d0d5de4

They are also compatible with many third party UIs and libraries - please see the list at the top of this README.

Explanation of quantisation methods

Click to see details

The new methods available are:

• GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
• GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
• GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
• GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
• GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw

Refer to the Provided Files table below to see what files use which methods, and how.

Example llama.cpp command

Make sure you are using llama.cpp from commit d0d5de4

./main -ngl 35 -m mixtral-8x22b-v0.1-instruct-oh-Q8_0-00001-of-00004.gguf --color -c 32768 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant"

Note that we are loading the GGUF of 00001-of-XXXXX. Using the latest llama.cpp will load the rest of the parts.

Change -ngl 32 to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.

Change -c 32768 to the desired sequence length. For extended sequence models - eg 8K, 16K, 32K - the necessary RoPE scaling parameters are read from the GGUF file and set by llama.cpp automatically. Note that longer sequence lengths require much more resources, so you may need to reduce this value.

If you want to have a chat-style conversation, replace the -p <PROMPT> argument with -i -ins

For other parameters and how to use them, please refer to the llama.cpp documentation

How to run in text-generation-webui

Further instructions can be found in the text-generation-webui documentation, here: text-generation-webui/docs/04 ‐ Model Tab.md.

How to run from Python code

You can use GGUF models from Python using the llama-cpp-python or ctransformers libraries. Note that at the time of writing (Nov 27th 2023), ctransformers has not been updated for some time and is not compatible with some recent models. Therefore I recommend you use llama-cpp-python.

How to load this model in Python code, using llama-cpp-python

For full documentation, please see: llama-cpp-python docs.

First install the package

Run one of the following commands, according to your system:

# Base ctransformers with no GPU acceleration
pip install llama-cpp-python
# With NVidia CUDA acceleration
CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python
# Or with OpenBLAS acceleration
CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS" pip install llama-cpp-python
# Or with CLBLast acceleration
CMAKE_ARGS="-DLLAMA_CLBLAST=on" pip install llama-cpp-python
# Or with AMD ROCm GPU acceleration (Linux only)
CMAKE_ARGS="-DLLAMA_HIPBLAS=on" pip install llama-cpp-python
# Or with Metal GPU acceleration for macOS systems only
CMAKE_ARGS="-DLLAMA_METAL=on" pip install llama-cpp-python

# In windows, to set the variables CMAKE_ARGS in PowerShell, follow this format; eg for NVidia CUDA:
$env:CMAKE_ARGS = "-DLLAMA_OPENBLAS=on"
pip install llama-cpp-python

Simple llama-cpp-python example code

from llama_cpp import Llama

# Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
llm = Llama(
  model_path="./mixtral-8x22b-v0.1-instruct-oh-Q8_0-00001-of-00004.gguf",  # Download the model file first
  n_ctx=32768,  # The max sequence length to use - note that longer sequence lengths require much more resources
  n_threads=8,            # The number of CPU threads to use, tailor to your system and the resulting performance
  n_gpu_layers=35         # The number of layers to offload to GPU, if you have GPU acceleration available
)

# Simple inference example
output = llm(
  "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant", # Prompt
  max_tokens=512,  # Generate up to 512 tokens
  stop=["</s>"],   # Example stop token - not necessarily correct for this specific model! Please check before using.
  echo=True        # Whether to echo the prompt
)

# Chat Completion API

llm = Llama(model_path="./mixtral-8x22b-v0.1-instruct-oh-Q8_0-00001-of-00004.gguf", chat_format="llama-2")  # Set chat_format according to the model you are using
llm.create_chat_completion(
    messages = [
        {"role": "system", "content": "You are a story writing assistant."},
        {
            "role": "user",
            "content": "Write a story about llamas."
        }
    ]
)

How to use with LangChain

Here are guides on using llama-cpp-python and ctransformers with LangChain:

• LangChain + llama-cpp-python
• LangChain + ctransformers

Original Model Card for Fireworks Mixtral 8x22b Instruct OH

Fireworks Mixtral 8x22b Instruct OH is an instruct version of the latest MoE model from mistralai - 8x22b. This model was finetuned on ~10K entries from OpenHermes dataset by NousResearch. Huge shoutout to Teknium and the NousResearch team for this high-quality SFT dataset.

Try out the model on the Fireworks platform.

How to Get Started with the Model

For trying out the model on a hosted platform go here.

The model uses Vicuna chat template:

USER: ...

ASSISTANT: ...

or:

SYSTEM: ...

USER: ...

ASSISTANT: ...

Training Data

To save training time, we sample & format 10K samples from OpenHermes 2.5 dataset. We perform reformatting so as to suit the vicuna chat template. Apart from that dataset is used as is.

Training Setup

The model was trained on 8xA100 [80G] using Deepspeed Zero 3. We had to fix an existing bug with deepspeed zero 3 where interaction between zero 3 & MoE causes training to get stuck.

Model Examination

Itinerary Question

SYSTEM: You are a helpful assistant. Please be kind, patient and truthful in answering the questions.

USER: Create a travel itinerary for the bahamas. Spare no detail.

ASSISTANT: Sure, I'd be happy to help you plan a trip to the Bahamas! Here's a detailed itinerary for your consideration:

Day 1: Arrival in Nassau

• Arrive at Lynden Pindling International Airport in Nassau, the capital city of the Bahamas.
• Check into your hotel or resort.
• Spend the afternoon exploring the city, visiting places like the Queen's Staircase, Fort Fincastle, and the Straw Market.
• Enjoy a delicious Bahamian dinner at one of the local restaurants.

Day 2: Nassau City Tour and Atlantis Resort

• Start your day with a guided city tour of Nassau, visiting historical sites like the Government House, Parliament Square, and the Pirates of Nassau Museum.
• In the afternoon, head to the famous Atlantis Resort on Paradise Island. Explore the resort's attractions, such as the Aquaventure Water Park, the Marine Habitat, and the casino.
• End the day with a relaxing dinner at one of the resort's restaurants.

Day 3: Exuma Cays Day Trip

• Take a day trip to the Exuma Cays, a group of over 365 islands and cays.
• Visit popular spots like the Swimming Pigs at Big Major Cay, the Thunderball Grotto (featured in the James Bond movie "Thunderball"), and the stunning sandbars.
• Enjoy a picnic lunch on one of the pristine beaches.
• Return to Nassau in the evening.

Day 4: Blue Lagoon Island and Beach Day

• Spend the morning at Blue Lagoon Island, a private island just a short boat ride from Nassau.
• Enjoy activities like swimming, snorkeling, kayaking, or simply relaxing on the beach.
• In the afternoon, head back to Nassau and spend some time at one of the city's beautiful beaches, such as Cable Beach or Junkanoo Beach.
• Have dinner at a local seafood restaurant.

Day 5: Departure

• Depending on your flight schedule, you may have some time to explore Nassau further or do some last-minute shopping before heading to the airport for your departure.

I hope this itinerary helps you plan your trip to the Bahamas! Let me know if you have any other questions or need further assistance.

QnA Question

SYSTEM: You are a helpful assistant. Please be kind, patient and truthful in answering the questions.

USER: What is the medical term for strong headache on one side of the head?

ASSISTANT: The medical term for a strong headache on one side of the head is called a migraine.

Knowledge Question

SYSTEM: You are a helpful assistant. Please be kind, patient and truthful in answering the questions.

USER: How can I get stock price of stripe?

ASSISTANT: Stripe is a private company and does not have publicly traded stock. Therefore, you cannot get the stock price for Stripe.