liodon-ai/phi-2-imatrix-GGUF overview
phi 2 — iMatrix GGUF GGUF quantizations of microsoft/phi 2 https://huggingface.co/microsoft/phi 2 , published by Liodon AI https://huggingface.co/liodon ai . Q…
Runs locally from ~968.5 MB disk (4 GB VRAM class GPUs with llama.cpp / guIDE).
Repository Files & Downloads
| File | Type | Quantization | Size | Link |
|---|---|---|---|---|
| phi-2-IQ2_M.gguf | GGUF | IQ2_M | 968.5 MB | Download |
| phi-2-IQ3_M.gguf | GGUF | IQ3_M | 1.23 GB | Download |
| phi-2-IQ4_XS.gguf | GGUF | IQ4_XS | 1.42 GB | Download |
| phi-2-Q4_K_M.gguf | GGUF | Q4_K_M | 1.62 GB | Download |
| phi-2-Q5_K_M.gguf | GGUF | Q5_K_M | 1.87 GB | Download |
| phi-2-Q6_K.gguf | GGUF | Q6_K | 2.13 GB | Download |
| phi-2-Q8_0.gguf | GGUF | Q8_0 | 2.75 GB | Download |
Model Details
| Model ID | liodon-ai/phi-2-imatrix-GGUF |
|---|---|
| Author | liodon-ai |
| Pipeline | text-generation |
| License | other |
| Base model | microsoft/phi-2 |
| Last modified | 2026-07-13T04:54:56.000Z |
Model README
---
license: other
base_model: microsoft/phi-2
base_model_relation: quantized
pipeline_tag: text-generation
library_name: gguf
tags:
- gguf
- ollama
- local-llm
- llama.cpp
- lm-studio
- quantized
- imatrix
- sub-4-bit
- phi
quantized_by: liodon-ai
---
phi-2 — iMatrix GGUF
GGUF quantizations of microsoft/phi-2, published by Liodon AI.
Quick Start
llama.cpp
llama-cli -hf liodon-ai/phi-2-imatrix-GGUF:Q4_K_MOllama
ollama run hf.co/liodon-ai/phi-2-imatrix-GGUF:Q4_K_MLM Studio / Jan — search liodon-ai/phi-2-imatrix-GGUF and pick your quant.
Quants
| Quant | Size | VRAM est. | Notes |
|-------|------|-----------|-------|
| IQ2_M | 1.02 GB | ~1 GB | 2-bit, iMatrix — smallest usable |
| IQ3_M | 1.32 GB | ~2 GB | 3-bit, iMatrix — great quality/size tradeoff |
| IQ4_XS | 1.52 GB | ~2 GB | 4-bit extra-small, iMatrix |
| Q4_K_M | 1.74 GB | ~2 GB | 4-bit, iMatrix-calibrated (recommended) |
| Q5_K_M | 2.00 GB | ~2 GB | 5-bit, iMatrix-calibrated |
| Q6_K | 2.29 GB | ~3 GB | 6-bit, iMatrix-calibrated, near-lossless |
| Q8_0 | 2.96 GB | ~3 GB | 8-bit, essentially lossless |
What is iMatrix?
Standard quantization treats all weights equally. iMatrix runs 128 calibration chunks through
the full-precision model to find which weights matter most, then allocates more precision where
it counts. At Q2/Q3/Q4 this means noticeably better coherence and instruction-following —
same file size, better output.
Calibration: 2M tokens of WikiText-103.
> Also see plain (non-iMatrix) quants: liodon-ai/phi-2-GGUF
Source
- Model: microsoft/phi-2
- License: other
---
Quantized by Liodon AI
Run liodon-ai/phi-2-imatrix-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models