Nllb 200 3.3B needs about 9.9 GB of memory at its main file (a 7.8 GB file) with an 8K context window, so it fits on a 12 GB GPU.
Nllb 200 3.3B is a 3.87B-parameter Nllb model in GGUF format, quantized by acceldium. Its trained context length is 1K tokens. The repository has 1 quant file, from 7.8 GB to 7.8 GB.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 7.8 GB | 8.5 GB | Too big | Get file → |
What GPU do you need for Nllb 200 3.3B?
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: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: the best quant that fits is the main file (7.8 GB file, about 9.9 GB in use).
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Nllb 200 3.3B
How much VRAM does Nllb 200 3.3B need?
About 9.9 GB at its main file with an 8K context window. That is the 7.8 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 Nllb 200 3.3B on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 9.9 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Nllb 200 3.3B GGUF?
It depends on the quant: file is 7.8 GB.
What context length does Nllb 200 3.3B support?
The GGUF header lists a trained context length of 1,024 tokens (1K). The KV cache grows with the context you use: about 1.6 GB at 8K and 0.2 GB at 1K.
Where can I download Nllb 200 3.3B GGUF?
The files are on Hugging Face in the acceldium/nllb-200-3.3B-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 Nllb 200 3.3B use?
Hugging Face lists the license as cc-by-nc-4.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.