BI

bigscience bloom 3b

BigscienceGGUF by RichardErkhovUpdated 2024-04-27
Download on Hugging Face
Parameters3.64B
Context3K
Layers30
Architecturebloom
Smallest quant1.4 GB
Quick answer

bigscience bloom 3b needs about 5.4 GB of memory at Q4_K_M (a 2.4 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 21 quants, from 1.4 GB (Q6_K) to 2.7 GB (Q5_K).

bigscience bloom 3b is a 3.64B-parameter Bigscience model in GGUF format, quantized by RichardErkhov. Its trained context length is 3K tokens. The repository has 21 quant files, from 1.4 GB (Q6_K) to 2.7 GB (Q5_K).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q6_K1.4 GB 2.7 GB
Fits Get file →
Q2_K1.6 GB 2.9 GB
Fits Get file →
Q5_11.7 GB 3.0 GB
Fits Get file →
Q5_K_M1.8 GB 3.1 GB
Fits Get file →
IQ3_XS1.8 GB 3.1 GB
Fits Get file →
IQ3_S1.8 GB 3.1 GB
Fits Get file →
Q3_K_S1.8 GB 3.1 GB
Fits Get file →
IQ3_M1.9 GB 3.2 GB
Fits Get file →
Q3_K2.0 GB 3.3 GB
Fits Get file →
Q3_K_M2.0 GB 3.3 GB
Fits Get file →
IQ4_XS2.2 GB 3.4 GB
Fits Get file →
Q3_K_L2.2 GB 3.5 GB
Fits Get file →
Q4_02.2 GB 3.5 GB
Fits Get file →
IQ4_NL2.2 GB 3.5 GB
Fits Get file →
Q4_K_S2.2 GB 3.5 GB
Fits Get file →
Q4_K2.4 GB 3.7 GB
Fits Get file →
Q4_K_M2.4 GB 3.7 GB
Fits Get file →
Q4_12.4 GB 3.7 GB
Fits Get file →
Q5_02.6 GB 3.9 GB
Fits Get file →
Q5_K_S2.6 GB 3.9 GB
Fits Get file →
Q5_K2.7 GB 4.0 GB
Fits Get file →

What GPU do you need for bigscience bloom 3b?

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: no quant fits at an 8K context.
  • 6 GB VRAM: the best quant that fits is Q5_K (2.7 GB file, about 5.8 GB in use).

Which bigscience bloom 3b quant should you choose?

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

If memory is tight, the smallest file is Q6_K at 1.4 GB, with the quality trade-offs that come with fewer bits. Very close to the original quality for most uses, at a clearly larger size than 4-bit.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturebloom
Parameters3.64B
Layers30
Trained context2,560 tokens
KV cache (fp16)300 KB per token

Source

Quantized byRichardErkhov
Downloads (HF)60
Last updated2024-04-27

Frequently asked questions about bigscience bloom 3b

How much VRAM does bigscience bloom 3b need?

About 5.4 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 bigscience bloom 3b on an 8 GB GPU?

Yes, with Q5_K: the 2.7 GB file needs about 5.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of bigscience bloom 3b GGUF?

It depends on the quant: Q6_K is 1.4 GB, Q2_K is 1.6 GB, Q5_1 is 1.7 GB, Q5_K_M is 1.8 GB, IQ3_XS is 1.8 GB, IQ3_S is 1.8 GB and 15 more.

What context length does bigscience bloom 3b support?

The GGUF header lists a trained context length of 2,560 tokens (3K). The KV cache grows with the context you use: about 2.5 GB at 8K and 0.8 GB at 3K.

Where can I download bigscience bloom 3b GGUF?

The files are on Hugging Face in the RichardErkhov/bigscience_-_bloom-3b-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 RichardErkhov

All by RichardErkhov →

More Bigscience models

All Bigscience →

Other small models

All small models →