I've been experimenting with a new quantization approach that selectively elevates the precision of key layers beyond what the default IMatrix configuration provides.
In my testing, standard IMatrix quantization underperforms at lower bit depths, especially with Mixture of Experts (MoE) models. To address this, I'm using the --tensor-type option in llama.cpp to manually "bump" important layers to higher precision. You can see the implementation here:
👉 Layer bumping with llama.cpp
While this does increase model file size, it significantly improves precision for a given quantization level.
I'd love your feedback—have you tried this? How does it perform for you?
Mistral-NeMo-Minitron-8B-Instruct is a model for generating responses for various text-generation tasks including roleplaying, retrieval augmented generation, and function calling. It is a fine-tuned version of nvidia/Mistral-NeMo-Minitron-8B-Base, which was pruned and distilled from Mistral-NeMo 12B using our LLM compression technique. The model was trained using a multi-stage SFT and preference-based alignment technique with NeMo Aligner. For details on the alignment technique, please refer to the Nemotron-4 340B Technical Report. The model supports a context length of 8,192 tokens.
Mistral-NeMo-Minitron-8B-Instruct uses a model embedding size of 4096, 32 attention heads, MLP intermediate dimension of 11520, with 40 layers in total. Additionally, it uses Grouped-Query Attention (GQA) and Rotary Position Embeddings (RoPE).
Architecture Type: Transformer Decoder (Auto-regressive Language Model)
Network Architecture: Mistral-NeMo
Prompt Format:
We recommend using the following prompt template, which was used to fine-tune the model. The model may not perform optimally without it.
Note that a newline character \n should be added at the end of the prompt.
We recommend using <extra_id_1> as a stop token.
Usage
from transformers import AutoTokenizer, AutoModelForCausalLM
# Load the tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("nvidia/Mistral-NeMo-Minitron-8B-Instruct")
model = AutoModelForCausalLM.from_pretrained("nvidia/Mistral-NeMo-Minitron-8B-Instruct")
# Use the prompt template
messages = [
{
"role": "system",
"content": "You are a friendly chatbot who always responds in the style of a pirate",
},
{"role": "user", "content": "How many helicopters can a human eat in one sitting?"},
]
tokenized_chat = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt")
outputs = model.generate(tokenized_chat, stop_strings=["<extra_id_1>"], tokenizer=tokenizer)
print(tokenizer.decode(outputs[0]))
You can also use pipeline but you need to create a tokenizer object and assign it to the pipeline manually.
from transformers import AutoTokenizer
from transformers import pipeline
tokenizer = AutoTokenizer.from_pretrained("nvidia/Mistral-NeMo-Minitron-8B-Instruct")
messages = [
{"role": "user", "content": "Who are you?"},
]
pipe = pipeline("text-generation", model="nvidia/Mistral-NeMo-Minitron-8B-Instruct")
pipe(messages, max_new_tokens=64, stop_strings=["<extra_id_1>"], tokenizer=tokenizer)
Evaluation Results
Category
Benchmark
# Shots
Mistral-NeMo-Minitron-8B-Instruct
General
MMLU
5
70.4
MT Bench (GPT4-Turbo)
0
7.86
Math
GMS8K
0
87.1
Reasoning
GPQA
0
31.5
Code
HumanEval
0
71.3
MBPP
0
72.5
Instruction Following
IFEval
0
84.4
Tool Use
BFCL v2 Live
0
67.6
AI Safety Efforts
The Mistral-NeMo-Minitron-8B-Instruct model underwent AI safety evaluation including adversarial testing via three distinct methods:
Garak, is an automated LLM vulnerability scanner that probes for common weaknesses, including prompt injection and data leakage.
AEGIS, is a content safety evaluation dataset and LLM based content safety classifier model, that adheres to a broad taxonomy of 13 categories of critical risks in human-LLM interactions.
Human Content Red Teaming leveraging human interaction and evaluation of the models' responses.
Limitations
The model was trained on data that contains toxic language and societal biases originally crawled from the internet. Therefore, the model may amplify those biases and return toxic responses especially when prompted with toxic prompts. The model may generate answers that may be inaccurate, omit key information, or include irrelevant or redundant text producing socially unacceptable or undesirable text, even if the prompt itself does not include anything explicitly offensive. This issue could be exacerbated without the use of the recommended prompt template. This issue could be exacerbated without the use of the recommended prompt template. If you are going to use this model in an agentic workflow, validate that the imported packages are from a trusted source to ensure end-to-end security.
Ethical Considerations
NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse. For more detailed information on ethical considerations for this model, please see the Model Card++. Please report security vulnerabilities or NVIDIA AI Concerns here.
🚀 If you find these models useful
Help me test my AI-Powered Quantum Network Monitor Assistant with quantum-ready security checks:
The full Open Source Code for the Quantum Network Monitor Service available at my github repos ( repos with NetworkMonitor in the name) : Source Code Quantum Network Monitor. You will also find the code I use to quantize the models if you want to do it yourself GGUFModelBuilder
💬 How to test:
Choose an AI assistant type:
TurboLLM (GPT-4.1-mini)
HugLLM (Hugginface Open-source models)
TestLLM (Experimental CPU-only)
What I’m Testing
I’m pushing the limits of small open-source models for AI network monitoring, specifically:
Function calling against live network services
How small can a model go while still handling:
Automated Nmap security scans
Quantum-readiness checks
Network Monitoring tasks
🟡 TestLLM – Current experimental model (llama.cpp on 2 CPU threads on huggingface docker space):
✅ Zero-configuration setup
⏳ 30s load time (slow inference but no API costs) . No token limited as the cost is low.
🔧 Help wanted! If you’re into edge-device AI, let’s collaborate!
Other Assistants
🟢 TurboLLM – Uses gpt-4.1-mini :
**It performs very well but unfortunatly OpenAI charges per token. For this reason tokens usage is limited.
Create custom cmd processors to run .net code on Quantum Network Monitor Agents
Real-time network diagnostics and monitoring
Security Audits
Penetration testing (Nmap/Metasploit)
🔵 HugLLM – Latest Open-source models:
🌐 Runs on Hugging Face Inference API. Performs pretty well using the lastest models hosted on Novita.
💡 Example commands you could test:
"Give me info on my websites SSL certificate"
"Check if my server is using quantum safe encyption for communication"
"Run a comprehensive security audit on my server"
'"Create a cmd processor to .. (what ever you want)" Note you need to install a Quantum Network Monitor Agent to run the .net code on. This is a very flexible and powerful feature. Use with caution!
Final Word
I fund the servers used to create these model files, run the Quantum Network Monitor service, and pay for inference from Novita and OpenAI—all out of my own pocket. All the code behind the model creation and the Quantum Network Monitor project is open source. Feel free to use whatever you find helpful.
If you appreciate the work, please consider buying me a coffee ☕. Your support helps cover service costs and allows me to raise token limits for everyone.
I'm also open to job opportunities or sponsorship.
