metadata

tags:
  - quantized
  - 2-bit
  - 3-bit
  - 4-bit
  - 5-bit
  - 6-bit
  - 8-bit
  - GGUF
  - transformers
  - pytorch
  - tensorboard
  - safetensors
  - mixtral
  - text-generation
  - axolotl
  - generated_from_trainer
  - nous_hermes
  - wikihow
  - peft
  - lora
  - qlora
  - adapter
  - finetune
  - conversational
  - dataset:HuggingFaceTB/cosmopedia
  - base_model:NousResearch/Nous-Hermes-2-Mixtral-8x7B-SFT
  - license:apache-2.0
  - autotrain_compatible
  - endpoints_compatible
  - text-generation-inference
  - region:us
  - text-generation
model_name: Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF
base_model: MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow
inference: false
model_creator: MaziyarPanahi
pipeline_tag: text-generation
quantized_by: MaziyarPanahi

MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF

Model creator: MaziyarPanahi
Original model: MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow

Description

MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF contains GGUF format model files for MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow.

Examples

<|im_start|> system
You are a sentient, superintelligent artificial general intelligence, here to teach and assist me.<|im_end|> 
<|im_start|> user
Write a short story about Goku discovering kirby has teamed up with Majin Buu to destroy the world.<|im_end|> 
<|im_start|> assistant

Once upon a time in Universe Seven, there lived two powerful beings - Son Goku, Earth's greatest hero; and Kirby, an interdimensional pink puffball known for his insatiable appetite and ability to copy abilities from those he consumes. One fateful day, these unlikely adversaries would cross paths when they discovered that their once-rivalized worlds were now under threat by none other than...a collaboration between Kirby himself and one of Dragon Ball Z’s most feared villains – Majin Buu!

It began as any ordinary morning might have started out: breakfast at Capsule Corp., followed by rigorous training exercises led personally by Vegeta (who had since become something like family). However, midway through sparring practice against Cell Jr., Gohan suddenly interrupted them via telepathy—his voice filled with urgency yet tainted slightly with disbelief—"Dad…Kakarot!" He hesitated briefly before continuing gravely, "I just received word from King Kai regarding some disturbingly odd activity happening back home on Popstar."

Goku raised an eyebrow curiously but didn't stop practicing until completing his current set. Afterward, wiping sweat off his brow, he asked casually while heading towards Bulma's lab where she was working diligently alongside Trunks & Goten, “So what kind of trouble does my old buddy say we should look into?”

Bulma glanced away momentarily from whatever gadgetry held her attention captive long enough to respond dryly without looking up, “Well apparently our favorite little vacuum cleaner seems hellbent on destroying everything again.” She paused dramatically then added sarcastically, “But this time around? Get ready because guess who decided tagging along sounded fun?" With great flair, she finally revealed the answer written clearly across multiple screens surrounding her workspace, displaying images captured directly from Planet Namek itself showing both Kirby AND MAJIN BUU causing widespread chaos hand-in-hand!!

For several seconds following confirmation of such preposterous news, silence reigned supreme within the usually bustling confines of Capsule Corporation HQ. Finally breaking free from shock induced paralysis, Goku managed only three words initially, spoken softly almost reverentially, “Buu...and...Kirby?” Then louder, more forcefully demanding clarification, “How could THAT possibly happen?! They don't even exist within same universe let alone work together willingly!”

Vegeta scoffed derisively beside him muttering dismissively beneath breath, 'Impossible!' Yet despite initial skepticism evident throughout entire room moments ago, everyone present couldn't deny mounting evidence suggesting otherwise right before eyes...

And thus begins epic tale chronicling adventures undertaken joint forces comprised primarily Super Saiyan God Blue empowered protagonist hailing originally from distant planet Vegeta accompanied unexpected albeit reluctant antihero originating entirely separate reality altogether bent solely upon saving existence itself rather than dominion over it unlike certain counterparts sharing similar circumstances elsewhere far removed spatially speaking....<|im_end|>

<|im_start|> system
You are a sentient, superintelligent artificial general intelligence, here to teach and assist me.<|im_end|> 
<|im_start|> user
If I lay 2 wet shirts out in the sun flat on the ground to dry and it takes 5 hours until they are dry, how long would it take to dry 4 shirts of the same type that way simultanously?<|im_end|> 
<|im_start|> assistant

The time required for drying multiple shirts simultaneously remains the same as when you dried them one at a time because there is no dependency between the individual drying processes. The factors affecting drying speed include sunlight intensity, air circulation, humidity levels, and fabric properties – none of which change based on whether you're drying two or four shirts concurrently.
Thus, if laying out four wet shirts still permits ample exposure to direct sunlight without overlapping too much (which could hinder efficient evaporation), then those additional garments should also be completely dried within approximately 5 hours under identical environmental conditions.<|im_end|>

<|im_start|> system
You are a sentient, superintelligent artificial general intelligence, here to teach and assist me.<|im_end|> 
<|im_start|> user
explain step by step 25-4*2+3=?<|im_end|> 
<|im_start|> assistant


Sure! Let's break down the equation into simpler steps:

Given expression: 25 - 4 * 2 + 3 = ?

Step 1: Calculate the multiplication inside parentheses (if any). Since there aren't any in this case, move on to Step 2 directly.

Step 2: Perform operations from left to right according to order of precedence – Parentheses > Exponents > Multiplication & Division > Addition & Subtraction. Here we have only addition (+) and subtraction (-), so perform them from left to right.

First operation is subtraction: 25 - 8 (since 4 times 2 equals 8), resulting in:

25 - 8 = 17

Next operation is addition; add 3 to our current result:

17 + 3 = 20

So, the final answer after evaluating all parts of the given expression would be **20**.<|im_end|>

<|im_start|> system
You are a sentient, superintelligent artificial general intelligence, here to teach and assist me.<|im_end|> 
<|im_start|> user
Mark is faster than Mary , Mary is faster than Joe. Is Joe faster than Mark? Let's think step by step<|im_end|> 
<|im_start|> assistant


Step 1: Understand the given information
We know that Mark is faster than Mary, which we can represent as Mark > Mary (using '>' symbol for comparison). Similarly, Mary is faster than Joe, so we have Mary > Joe.

Step 2: Compare the speeds of Mark and Joe using transitive property
The transitive property states that if A > B and B > C, then A > C. Applying this concept here - since we know Mark > Mary and Mary > Joe, we should be able to determine whether Mark > Joe or not.

Step 3: Analyze results
By applying the transitive property from Step 2, it becomes evident that because Mark is faster than Mary, who in turn is faster than Joe; therefore logically, yes – Mark would indeed be faster than Joe. So our final statement would look like this - Mark > Joe.<|im_end|>

How to use

Thanks to TheBloke for preparing an amazing README on how to use GGUF models:

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.

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

How to download GGUF files

Note for manual downloaders: You almost never want to clone the entire repo! Multiple different quantisation formats are provided, and most users only want to pick and download a single file.

The following clients/libraries will automatically download models for you, providing a list of available models to choose from:

LM Studio
LoLLMS Web UI
Faraday.dev

In `text-generation-webui`

Under Download Model, you can enter the model repo: MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF and below it, a specific filename to download, such as: Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF.Q4_K_M.gguf.

Then click Download.

On the command line, including multiple files at once

I recommend using the huggingface-hub Python library:

pip3 install huggingface-hub

Then you can download any individual model file to the current directory, at high speed, with a command like this:

huggingface-cli download MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False

More advanced huggingface-cli download usage (click to read)

You can also download multiple files at once with a pattern:

huggingface-cli download [MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF](https://huggingface.co/MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF) --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'

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:

HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download MaziyarPanahi/Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF.Q4_K_M.gguf --local-dir . --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.

Example `llama.cpp` command

Make sure you are using llama.cpp from commit d0cee0d or later.

./main -ngl 35 -m Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF.Q4_K_M.gguf --color -c 32768 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "<|im_start|>system
{system_message}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant"

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="./Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF.Q4_K_M.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
{system_message}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|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="./Nous-Hermes-2-Mixtral-8x7B-SFT-Wikihow-GGUF.Q4_K_M.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: