CelesteImperia/Phi-3.5-mini-instruct-Platinum-GGUF

Phi-3.5-mini-instruct-GGUF (Platinum Series)

This repository contains the Platinum Series universal GGUF release of Phi-3.5-mini-instruct. This collection provides multiple quantization levels optimized for cross-platform performance, offering advanced reasoning capabilities with 128k context support.

📦 Available Files & Quantization Details

File Name	Quantization	Size	Accuracy	Recommended For
Phi-3.5-mini-instruct-Platinum-F16.gguf	FP16	~7.6 GB	100%	Master Reference / Benchmarking
Phi-3.5-mini-instruct-Platinum-Q8_0.gguf	Q8_0	~4.1 GB	99.9%	Platinum Reference / High-Fidelity
Phi-3.5-mini-instruct-Platinum-Q6_K.gguf	Q6_K	~3.1 GB	99.8%	High-Quality Inference
Phi-3.5-mini-instruct-Platinum-Q5_K_M.gguf	Q5_K_M	~2.8 GB	99.4%	Balanced Desktop Performance
Phi-3.5-mini-instruct-Platinum-Q4_K_M.gguf	Q4_K_M	~2.4 GB	98.8%	Mobile / Low-Power Efficiency

---

🐍 Python Inference (llama-cpp-python)

To run these engines using Python:

from llama_cpp import Llama

llm = Llama(
    model_path="Phi-3.5-mini-instruct-Platinum-Q8_0.gguf",
    n_gpu_layers=-1, # Target all layers to NVIDIA/Apple GPU
    n_ctx=4096
)

output = llm("Discuss the architectural benefits of Phi-3.5.", max_tokens=150)
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("Phi-3.5-mini-instruct-Platinum-Q8_0.gguf") {
    ContextSize = 4096,
    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: Phi-3.5
• Hardware Validation: Dual-GPU (RTX 3090 + RTX A4000)

---

☕ Support the Forge

Maintaining the production line for high-fidelity models requires significant hardware resources. If these tools power your research or industrial projects, 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