MI

MiniChat 2 3B iMat

MinichatGGUF by cgusUpdated 2024-05-17
Download on Hugging Face
Parameters3.02B
Context4K
Layers24
Architecturellama
Smallest quant1.0 GB
Quick answer

MiniChat 2 3B iMat needs about 4.8 GB of memory at Q4_K_M (a 1.8 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 9 quants, from 1.0 GB (IQ2_XS) to 3.2 GB (Q8_0).

MiniChat 2 3B iMat is a 3.02B-parameter Minichat model in GGUF format, quantized by cgus. Its trained context length is 4K tokens. The repository has 9 quant files, from 1.0 GB (IQ2_XS) to 3.2 GB (Q8_0).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ2_XS1.0 GB 2.7 GB
Fits Get file →
IQ2_M1.1 GB 2.8 GB
Fits Get file →
IQ3_M1.4 GB 3.1 GB
Fits Get file →
IQ4_XS1.6 GB 3.4 GB
Fits Get file →
IQ4_NL1.7 GB 3.4 GB
Fits Get file →
Q4_K_M1.8 GB 3.6 GB
Fits Get file →
Q5_K_M2.2 GB 3.9 GB
Fits Get file →
Q6_K2.5 GB 4.2 GB
Fits Get file →
Q8_03.2 GB 4.9 GB
Fits Get file →

What GPU do you need for MiniChat 2 3B iMat?

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 IQ2_M (1.1 GB file, about 4.0 GB in use).
  • 6 GB VRAM: the best quant that fits is Q6_K (2.5 GB file, about 5.4 GB in use).
  • 8 GB VRAM: the best quant that fits is Q8_0 (3.2 GB file, about 6.1 GB in use).

Which MiniChat 2 3B iMat quant should you choose?

Q4_K_M is the usual starting point for MiniChat 2 3B iMat: it balances file size and quality well. Here it is a 1.8 GB file.

The Q8_0 file is 3.2 GB, about 1.7 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 IQ2_XS at 1.0 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturellama
Parameters3.02B
Layers24
Trained context4,096 tokens
KV cache (fp16)288 KB per token

Source

Quantized bycgus
FamilyMinichat
Licenseapache-2.0
Downloads (HF)36
Last updated2024-05-17

Frequently asked questions about MiniChat 2 3B iMat

How much VRAM does MiniChat 2 3B iMat need?

About 4.8 GB at Q4_K_M with an 8K context window. That is the 1.8 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 MiniChat 2 3B iMat on an 8 GB GPU?

Yes, with Q8_0: the 3.2 GB file needs about 6.1 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of MiniChat 2 3B iMat GGUF?

It depends on the quant: IQ2_XS is 1.0 GB, IQ2_M is 1.1 GB, IQ3_M is 1.4 GB, IQ4_XS is 1.6 GB, IQ4_NL is 1.7 GB, Q4_K_M is 1.8 GB and 3 more.

What context length does MiniChat 2 3B iMat support?

The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 2.4 GB at 8K and 1.2 GB at 4K.

Where can I download MiniChat 2 3B iMat GGUF?

The files are on Hugging Face in the cgus/MiniChat-2-3B-iMat-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does MiniChat 2 3B iMat use?

Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Other GGUFs of MiniChat-2-3B

Similar models

Browse all →

More from cgus

All by cgus →

More Minichat models

All Minichat →

Other tiny models

All tiny models →