base_model, library_name, pipeline_tag, license, tags

base_model	library_name	pipeline_tag	license	tags
google/gemma-2-2b-it	gguf	text-generation	gemma	gguf llama-cpp gemma-2 celeste-imperia

Gemma-2-2B-IT-GGUF (Platinum Series)

This repository contains the Universal GGUF release of Gemma-2-2B-Instruct. This collection provides multiple quantization levels to support everything from high-VRAM workstations to mobile and edge devices.

📦 Available Files & Quantization Details

File Name	Quantization	Size	Accuracy	Recommended For
Gemma-2-2B-IT-Platinum-F16.gguf	FP16	~5.2 GB	100%	Master Reference / Benchmarking
Gemma-2-2B-IT-Platinum-Q8_0.gguf	Q8_0	~2.7 GB	99.9%	Platinum Reference / High-Fidelity
Gemma-2-2B-IT-Platinum-Q6_K.gguf	Q6_K	~2.1 GB	99.5%	High-end GPU / Complex Reasoning
Gemma-2-2B-IT-Platinum-Q5_K_M.gguf	Q5_K_M	~1.8 GB	99.0%	Balanced Desktop Performance
Gemma-2-2B-IT-Platinum-Q4_K_M.gguf	Q4_K_M	~1.5 GB	98.2%	Mobile / Low-VRAM / Efficiency

---

🐍 Python Inference (llama-cpp-python)

To run these engines using Python:

from llama_cpp import Llama

llm = Llama(
    model_path="Gemma-2-2B-IT-Platinum-Q8_0.gguf",
    n_gpu_layers=-1, # Target all layers to NVIDIA/Apple GPU
    n_ctx=2048
)

output = llm("Explain the architecture of Gemma 2.", 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("Gemma-2-2B-IT-Platinum-Q8_0.gguf") {
    ContextSize = 2048,
    GpuLayerCount = 35 
};

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

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

---

🏗️ Technical Details

• Optimization Tool: llama.cpp (CUDA-accelerated)
• Architecture: Gemma-2
• Hardware Validation: RTX 3090 + RTX A4000

---

☕ Support the Forge

Maintaining a high-capacity local AI warehouse with full-fidelity weights requires significant hardware resources. If these models power your industrial projects or 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