Mistral Nemo Instruct 2407 - llamafile

• Model creator: Mistral AI
• Original model: mistralai/Mistral-Nemo-Instruct-2407

The model is packaged into executable weights, which we call llamafiles. This makes it easy to use the model on Linux, MacOS, Windows, FreeBSD, OpenBSD, and NetBSD for AMD64 and ARM64.

Quickstart

Running the following on a desktop OS will launch a tab in your web browser with a chatbot interface.

wget https://huggingface.co/Mozilla/Mistral-Nemo-Instruct-2407-llamafile/resolve/main/Mistral-Nemo-Instruct-2407.Q6_K.llamafile
chmod +x Mistral-Nemo-Instruct-2407.Q6_K.llamafile
./Mistral-Nemo-Instruct-2407.Q6_K.llamafile

You then need to fill out the prompt / history template (see below).

This model has a max context window size of 128k tokens. By default, a context window size of 8192 tokens is used. You may increase this to the maximum by passing the -c 0 flag.

On GPUs with sufficient RAM, the -ngl 999 flag may be passed to use the system's NVIDIA or AMD GPU(s). On Windows, only the graphics card driver needs to be installed. If the prebuilt DSOs should fail, the CUDA or ROCm SDKs may need to be installed, in which case llamafile builds a native module just for your system.

For further information, please see the llamafile README.

Having trouble? See the "Gotchas" section of the README.

Prompting

To have a good working chat experience when using the web GUI, you need to fill out the text fields with the following values.

Prompt template:

{{prompt}}

{{history}}
{{char}}:

History template:

{{name}}: {{message}}

Here's an example of how to prompt Mistral on the command line:

./Mistral-Nemo-Instruct-2407.Q6_K.llamafile -p '[INST]The Belobog Academy has discovered a new, invasive species of algae that can double itself in one day, and in 30 days fills a whole reservoir - contaminating the water supply. How many days would it take for the algae to fill half of the reservoir?[/INST]'

About llamafile

llamafile is a new format introduced by Mozilla Ocho on Nov 20th 2023. It uses Cosmopolitan Libc to turn LLM weights into runnable llama.cpp binaries that run on the stock installs of six OSes for both ARM64 and AMD64.

---

Model Card for Mistral-Nemo-Instruct-2407

The Mistral-Nemo-Instruct-2407 Large Language Model (LLM) is an instruct fine-tuned version of the Mistral-Nemo-Base-2407. Trained jointly by Mistral AI and NVIDIA, it significantly outperforms existing models smaller or similar in size.

For more details about this model please refer to our release blog post.

Key features

• Released under the Apache 2 License
• Pre-trained and instructed versions
• Trained with a 128k context window
• Trained on a large proportion of multilingual and code data
• Drop-in replacement of Mistral 7B

Model Architecture

Mistral Nemo is a transformer model, with the following architecture choices:

• Layers: 40
• Dim: 5,120
• Head dim: 128
• Hidden dim: 14,336
• Activation Function: SwiGLU
• Number of heads: 32
• Number of kv-heads: 8 (GQA)
• Vocabulary size: 2**17 ~= 128k
• Rotary embeddings (theta = 1M)

Metrics

Main Benchmarks

Multilingual Benchmarks (MMLU)

Usage

The model can be used with three different frameworks

• mistral_inference: See here
• transformers: See here
• NeMo: See nvidia/Mistral-NeMo-12B-Instruct

Mistral Inference

Install

It is recommended to use mistralai/Mistral-Nemo-Instruct-2407 with mistral-inference. For HF transformers code snippets, please keep scrolling.

pip install mistral_inference

Download

from huggingface_hub import snapshot_download
from pathlib import Path

mistral_models_path = Path.home().joinpath('mistral_models', 'Nemo-Instruct')
mistral_models_path.mkdir(parents=True, exist_ok=True)

snapshot_download(repo_id="mistralai/Mistral-Nemo-Instruct-2407", allow_patterns=["params.json", "consolidated.safetensors", "tekken.json"], local_dir=mistral_models_path)

Chat

After installing mistral_inference, a mistral-chat CLI command should be available in your environment. You can chat with the model using

mistral-chat $HOME/mistral_models/Nemo-Instruct --instruct --max_tokens 256 --temperature 0.35

E.g. Try out something like:

How expensive would it be to ask a window cleaner to clean all windows in Paris. Make a reasonable guess in US Dollar.

Instruct following

from mistral_inference.transformer import Transformer
from mistral_inference.generate import generate

from mistral_common.tokens.tokenizers.mistral import MistralTokenizer
from mistral_common.protocol.instruct.messages import UserMessage
from mistral_common.protocol.instruct.request import ChatCompletionRequest

tokenizer = MistralTokenizer.from_file(f"{mistral_models_path}/tekken.json")
model = Transformer.from_folder(mistral_models_path)

prompt = "How expensive would it be to ask a window cleaner to clean all windows in Paris. Make a reasonable guess in US Dollar."

completion_request = ChatCompletionRequest(messages=[UserMessage(content=prompt)])

tokens = tokenizer.encode_chat_completion(completion_request).tokens

out_tokens, _ = generate([tokens], model, max_tokens=64, temperature=0.35, eos_id=tokenizer.instruct_tokenizer.tokenizer.eos_id)
result = tokenizer.decode(out_tokens[0])

print(result)

Function calling

from mistral_common.protocol.instruct.tool_calls import Function, Tool
from mistral_inference.transformer import Transformer
from mistral_inference.generate import generate

from mistral_common.tokens.tokenizers.mistral import MistralTokenizer
from mistral_common.protocol.instruct.messages import UserMessage
from mistral_common.protocol.instruct.request import ChatCompletionRequest


tokenizer = MistralTokenizer.from_file(f"{mistral_models_path}/tekken.json")
model = Transformer.from_folder(mistral_models_path)

completion_request = ChatCompletionRequest(
    tools=[
        Tool(
            function=Function(
                name="get_current_weather",
                description="Get the current weather",
                parameters={
                    "type": "object",
                    "properties": {
                        "location": {
                            "type": "string",
                            "description": "The city and state, e.g. San Francisco, CA",
                        },
                        "format": {
                            "type": "string",
                            "enum": ["celsius", "fahrenheit"],
                            "description": "The temperature unit to use. Infer this from the users location.",
                        },
                    },
                    "required": ["location", "format"],
                },
            )
        )
    ],
    messages=[
        UserMessage(content="What's the weather like today in Paris?"),
        ],
)

tokens = tokenizer.encode_chat_completion(completion_request).tokens

out_tokens, _ = generate([tokens], model, max_tokens=256, temperature=0.35, eos_id=tokenizer.instruct_tokenizer.tokenizer.eos_id)
result = tokenizer.decode(out_tokens[0])

print(result)

Transformers

[!IMPORTANT] NOTE: Until a new release has been made, you need to install transformers from source:

pip install git+https://github.com/huggingface/transformers.git

If you want to use Hugging Face transformers to generate text, you can do something like this.

from transformers import pipeline

messages = [
    {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"},
    {"role": "user", "content": "Who are you?"},
]
chatbot = pipeline("text-generation", model="mistralai/Mistral-Nemo-Instruct-2407",max_new_tokens=128)
chatbot(messages)

Function calling with transformers

To use this example, you'll need transformers version 4.42.0 or higher. Please see the function calling guide in the transformers docs for more information.

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "mistralai/Mistral-Nemo-Instruct-2407"
tokenizer = AutoTokenizer.from_pretrained(model_id)

def get_current_weather(location: str, format: str):
    """
    Get the current weather

    Args:
        location: The city and state, e.g. San Francisco, CA
        format: The temperature unit to use. Infer this from the users location. (choices: ["celsius", "fahrenheit"])
    """
    pass

conversation = [{"role": "user", "content": "What's the weather like in Paris?"}]
tools = [get_current_weather]

# render the tool use prompt as a string:
tool_use_prompt = tokenizer.apply_chat_template(
            conversation,
            tools=tools,
            tokenize=False,
            add_generation_prompt=True,
)

inputs = tokenizer(tool_use_prompt, return_tensors="pt")

model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto")

outputs = model.generate(**inputs, max_new_tokens=1000)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Note that, for reasons of space, this example does not show a complete cycle of calling a tool and adding the tool call and tool results to the chat history so that the model can use them in its next generation. For a full tool calling example, please see the function calling guide, and note that Mistral does use tool call IDs, so these must be included in your tool calls and tool results. They should be exactly 9 alphanumeric characters.

[!TIP] Unlike previous Mistral models, Mistral Nemo requires smaller temperatures. We recommend to use a temperature of 0.3.

Limitations

The Mistral Nemo Instruct model is a quick demonstration that the base model can be easily fine-tuned to achieve compelling performance. It does not have any moderation mechanisms. We're looking forward to engaging with the community on ways to make the model finely respect guardrails, allowing for deployment in environments requiring moderated outputs.

The Mistral AI Team

Albert Jiang, Alexandre Sablayrolles, Alexis Tacnet, Alok Kothari, Antoine Roux, Arthur Mensch, Audrey Herblin-Stoop, Augustin Garreau, Austin Birky, Bam4d, Baptiste Bout, Baudouin de Monicault, Blanche Savary, Carole Rambaud, Caroline Feldman, Devendra Singh Chaplot, Diego de las Casas, Eleonore Arcelin, Emma Bou Hanna, Etienne Metzger, Gaspard Blanchet, Gianna Lengyel, Guillaume Bour, Guillaume Lample, Harizo Rajaona, Henri Roussez, Hichem Sattouf, Ian Mack, Jean-Malo Delignon, Jessica Chudnovsky, Justus Murke, Kartik Khandelwal, Lawrence Stewart, Louis Martin, Louis Ternon, Lucile Saulnier, Lélio Renard Lavaud, Margaret Jennings, Marie Pellat, Marie Torelli, Marie-Anne Lachaux, Marjorie Janiewicz, Mickaël Seznec, Nicolas Schuhl, Niklas Muhs, Olivier de Garrigues, Patrick von Platen, Paul Jacob, Pauline Buche, Pavan Kumar Reddy, Perry Savas, Pierre Stock, Romain Sauvestre, Sagar Vaze, Sandeep Subramanian, Saurabh Garg, Sophia Yang, Szymon Antoniak, Teven Le Scao, Thibault Schueller, Thibaut Lavril, Thomas Wang, Théophile Gervet, Timothée Lacroix, Valera Nemychnikova, Wendy Shang, William El Sayed, William Marshall