DeepSeek-R1-Distill-Qwen-1.5B — iMatrix GGUF
GGUF quantizations of deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B, published by Liodon AI.
Quick Start
llama.cpp
llama-cli -hf liodon-ai/DeepSeek-R1-Distill-Qwen-1.5B-imatrix-GGUF:Q4_K_M
Ollama
ollama run hf.co/liodon-ai/DeepSeek-R1-Distill-Qwen-1.5B-imatrix-GGUF:Q4_K_M
LM Studio / Jan — search liodon-ai/DeepSeek-R1-Distill-Qwen-1.5B-imatrix-GGUF and pick your quant.
Quants
| Quant | Size | VRAM est. | Notes |
|---|---|---|---|
IQ2_M | 0.70 GB | ~1 GB | 2-bit, iMatrix — smallest usable |
IQ3_M | 0.88 GB | ~1 GB | 3-bit, iMatrix — great quality/size tradeoff |
IQ4_XS | 1.02 GB | ~1 GB | 4-bit extra-small, iMatrix |
Q4_K_M | 1.12 GB | ~1 GB | 4-bit, iMatrix-calibrated (recommended) |
Q5_K_M | 1.29 GB | ~1 GB | 5-bit, iMatrix-calibrated |
Q6_K | 1.46 GB | ~2 GB | 6-bit, iMatrix-calibrated, near-lossless |
Q8_0 | 1.89 GB | ~2 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/DeepSeek-R1-Distill-Qwen-1.5B-GGUF
Source
- Model: deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
- License: other
Quantized by Liodon AI