CelesteImperia/Mistral-Nemo-Instruct-Platinum-GGUF

Mistral-Nemo-12B-Instruct-v1-GGUF (Platinum Series)

This repository contains the Platinum Series universal GGUF release of Mistral-Nemo-12B-Instruct-v1. This collection provides multiple quantization levels optimized for high-performance reasoning and creative applications, bridging the gap between small and large-scale language models.

📦 Available Files & Quantization Details

File Name	Quantization	Size	Accuracy	Recommended For
Mistral-Nemo-12B-Instruct-Platinum-F16.gguf	FP16	~24.5 GB	100%	Master Reference / Benchmarking
Mistral-Nemo-12B-Instruct-Platinum-Q8_0.gguf	Q8_0	~13.0 GB	99.9%	Platinum Reference / High-Fidelity
Mistral-Nemo-12B-Instruct-Platinum-Q6_K.gguf	Q6_K	~10.0 GB	99.8%	High-End GPU / Complex Logic
Mistral-Nemo-12B-Instruct-Platinum-Q5_K_M.gguf	Q5_K_M	~8.7 GB	99.6%	Balanced Performance
Mistral-Nemo-12B-Instruct-Platinum-Q4_K_M.gguf	Q4_K_M	~7.4 GB	99.2%	Efficiency / Mid-Range VRAM

---

🐍 Python Inference (llama-cpp-python)

To run these engines using Python:

from llama_cpp import Llama

llm = Llama(
    model_path="Mistral-Nemo-12B-Instruct-Platinum-Q8_0.gguf",
    n_gpu_layers=-1, # Target all layers to NVIDIA/Apple GPU
    n_ctx=8192
)

output = llm("Discuss the impact of the Mistral-Nemo architecture.", max_tokens=200)
print(output["choices"][0]["text"])

---

💻 For C# / .NET Users (LLamaSharp)

This collection is fully compatible with .NET applications via the LLamaSharp library.

using LLama.Common;
using LLama;

var parameters = new ModelParams("Mistral-Nemo-12B-Instruct-Platinum-Q8_0.gguf") {
    ContextSize = 8192,
    GpuLayerCount = 40 
};

using var model = LLamaWeights.LoadFromFile(parameters);
using var context = model.CreateContext(parameters);
var executor = new InteractiveExecutor(context);

Console.WriteLine("Universal Logic Engine Active.");

---

🏗️ Technical Details

• Optimization Tool: llama.cpp (CUDA-accelerated)
• Architecture: Mistral Nemo (12B)
• Hardware Validation: Dual-GPU (RTX 3090 + RTX A4000)

---

☕ Support the Forge

Maintaining high-capacity workstations for model conversion requires hardware investment. If these tools power your industrial workflows or local research, please consider supporting the development:

Platform	Support Link
Global & India	Support via Razorpay

Scan to support via UPI (India Only):

---

Connect with the architect: Abhishek Jaiswal on LinkedIn