PH

phi 3

PhiGGUF by BaunRoboticsUpdated 2024-06-04
Download on Hugging Face
Parameters3.82B
Context4K
Layers32
Architecturephi3
Smallest quant0.0 GB
Quick answer

phi 3 needs about 6.1 GB of memory at Q4_K_M (a 2.4 GB file) with an 8K context window, so it fits on a 8 GB GPU. This repository has 15 quants, from 0.0 GB (Q8_0) to 7.6 GB.

phi 3 is a 3.82B-parameter Phi model in GGUF format, quantized by BaunRobotics. Its trained context length is 4K tokens. The repository has 15 quant files, from 1.4 GB (Q2_K) to 7.6 GB.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q8_00.0 GB 2.1 GB
Fits Get file →
Q2_K1.4 GB 3.5 GB
Fits Get file →
Q3_K_S1.7 GB 3.8 GB
Fits Get file →
Q3_K2.0 GB 4.1 GB
Fits Get file →
Q3_K_M2.0 GB 4.1 GB
Fits Get file →
Q4_02.2 GB 4.3 GB
Fits Get file →
Q4_K_M2.4 GB 4.5 GB
Fits Get file →
Q4_12.4 GB 4.5 GB
Fits Get file →
Q5_02.6 GB 4.8 GB
Fits Get file →
Q5_K_S2.6 GB 4.8 GB
Fits Get file →
Q6_K2.8 GB 4.9 GB
Fits Get file →
Q5_K2.8 GB 4.9 GB
Fits Get file →
Q5_K_M2.8 GB 4.9 GB
Fits Get file →
Q5_12.9 GB 5.0 GB
Fits Get file →
—7.6 GB 9.8 GB
Too big Get file →

What GPU do you need for phi 3?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: the best quant that fits is Q8_0 (0.0 GB file, about 3.7 GB in use).
  • 6 GB VRAM: the best quant that fits is Q4_0 (2.2 GB file, about 5.9 GB in use).
  • 8 GB VRAM: the best quant that fits is Q5_1 (2.9 GB file, about 6.6 GB in use).
  • 12 GB VRAM: the best quant that fits is the main file (7.6 GB file, about 11.4 GB in use).

Which phi 3 quant should you choose?

Q4_K_M is the usual starting point for phi 3: it balances file size and quality well. Here it is a 2.4 GB file.

The Q8_0 file is 0.0 GB, about 0.0 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is Q8_0 at 0.0 GB, with the quality trade-offs that come with fewer bits. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturephi3
Parameters3.82B
Layers32
Trained context4,096 tokens
KV cache (fp16)384 KB per token

Source

Quantized byBaunRobotics
FamilyPhi
Downloads (HF)54
Last updated2024-06-04

Frequently asked questions about phi 3

How much VRAM does phi 3 need?

About 6.1 GB at Q4_K_M with an 8K context window. That is the 2.4 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run phi 3 on an 8 GB GPU?

Yes, with Q5_1: the 2.9 GB file needs about 6.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of phi 3 GGUF?

It depends on the quant: Q8_0 is 0.0 GB, Q2_K is 1.4 GB, Q3_K_S is 1.7 GB, Q3_K is 2.0 GB, Q3_K_M is 2.0 GB, Q4_0 is 2.2 GB and 9 more.

What context length does phi 3 support?

The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 3.2 GB at 8K and 1.6 GB at 4K.

Where can I download phi 3 GGUF?

The files are on Hugging Face in the BaunRobotics/phi-3-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

Similar models

Browse all →

More from BaunRobotics

All by BaunRobotics →

More Phi models

All Phi →

Other small models

All small models →