MI

Minitron 4B Base

MinitronGGUF by legraphistaUpdated 2024-08-17
Download on Hugging Face
Parameters4.19B
Context4K
Layers32
Architecturenemotron
Smallest quant1.9 GB
Quick answer

Minitron 4B Base needs about 4.2 GB of memory at IQ4_NL (a 2.6 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 17 quants, from 1.9 GB (Q2_K) to 8.4 GB.

Minitron 4B Base – a 4.19B parameter transformer decoder

Minitron 4B Base is a pruned and distilled version of Nemotron-4 15B. It contains 4.19B parameters and was trained with 94B tokens. The model supports text input and output only.

Good at

  • Text generation
  • Multilingual support
  • Code understanding

Limits

  • No vision support
  • License is NVIDIA Open Model License
  • Requires 8k context for full performance

Best for

  • Research prototypes
  • Low‑cost inference on GPUs
  • Fine‑tuning for niche domains

Who it’s for. Researchers and developers who need a compact, transformer‑based LLM for text tasks. The model is released under the NVIDIA Open Model License and is intended for research and development only.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K1.9 GB 2.9 GB
Fits Get file →
IQ3_XS2.1 GB 3.1 GB
Fits Get file →
IQ3_S2.1 GB 3.2 GB
Fits Get file →
Q3_K_S2.1 GB 3.2 GB
Fits Get file →
IQ3_M2.2 GB 3.2 GB
Fits Get file →
Q3_K2.3 GB 3.3 GB
Fits Get file →
Q3_K_L2.5 GB 3.5 GB
Fits Get file →
IQ4_XS2.5 GB 3.5 GB
Fits Get file →
IQ4_NL2.6 GB 3.6 GB
Fits Get file →
Q4_K_S2.6 GB 3.6 GB
Fits Get file →
Q4_K2.7 GB 3.7 GB
Fits Get file →
Q5_K_S3.0 GB 4.0 GB
Fits Get file →
Q5_K3.1 GB 4.1 GB
Fits Get file →
Q6_K3.4 GB 4.5 GB
Fits Get file →
Q8_04.5 GB 5.5 GB
Fits Get file →
BF168.4 GB 9.4 GB
Too big Get file →
—8.4 GB 9.4 GB
Too big Get file →

For 8 GB VRAM, use Q4_K or Q4_K_S (file size 2.58 GB, memory 4.2 GB). For 12 GB VRAM, Q5_K or Q5_K_S (file size 3.06 GB, memory 4.6 GB). For 16 GB VRAM, Q6_K (file size 3.45 GB, memory 5 GB).

What GPU do you need for Minitron 4B Base?

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 Q3_K_L (2.5 GB file, about 4.0 GB in use).
  • 6 GB VRAM: the best quant that fits is Q8_0 (4.5 GB file, about 6.0 GB in use).
  • 8 GB VRAM: the best quant that fits is Q8_0 (4.5 GB file, about 6.0 GB in use).
  • 12 GB VRAM: the best quant that fits is BF16 (8.4 GB file, about 10.0 GB in use).

Which Minitron 4B Base quant should you choose?

IQ4_NL (2.6 GB) is the middle option in this repository. The usual sweet spot between file size and quality, and the quant most people start with.

If memory is tight, the smallest file is Q2_K at 1.9 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

Architecturenemotron
Parameters4.19B
Layers32
Trained context4,096 tokens
KV cache (fp16)128 KB per token

Source

Quantized bylegraphista
FamilyMinitron
Licenseother
Downloads (HF)1,853
Last updated2024-08-17

Frequently asked questions about Minitron 4B Base

How much VRAM does Minitron 4B Base need?

About 4.2 GB at IQ4_NL with an 8K context window. That is the 2.6 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 Minitron 4B Base on an 8 GB GPU?

Yes, with Q8_0: the 4.5 GB file needs about 6.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Minitron 4B Base GGUF?

It depends on the quant: Q2_K is 1.9 GB, IQ3_XS is 2.1 GB, IQ3_S is 2.1 GB, Q3_K_S is 2.1 GB, IQ3_M is 2.2 GB, Q3_K is 2.3 GB and 11 more.

What context length does Minitron 4B Base support?

The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 1.1 GB at 8K and 0.5 GB at 4K.

Where can I download Minitron 4B Base GGUF?

The files are on Hugging Face in the legraphista/Minitron-4B-Base-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 Minitron 4B Base use?

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

Can I run Minitron 4B Base on 8 GB VRAM?

Yes, use the Q4_K or Q4_K_S quantization which requires about 4.2 GB of VRAM at 8k context.

Does the model support image input?

No, the model only accepts text input and produces text output.

What license applies to this model?

The model is released under the NVIDIA Open Model License Agreement.

Other GGUFs of Minitron-4B-Base

Similar models

Browse all →

More from legraphista

All by legraphista →

More Minitron models

All Minitron →

Other small models

All small models →