What are the hardware requirements for Llama 3.1 Minitron 4B Width Base?

Llama 3.1 Minitron 4B Width Base requires approximately 9 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Llama 3.1 Minitron 4B Width Base and how large is it?

Llama 3.1 Minitron 4B Width Base is developed by nvidia, a model with 4b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Llama 3.1 Minitron 4B Width Base?

Llama 3.1 Minitron 4B Width Base is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Llama 3.1 Minitron 4B Width Base by nvidia — VRAM 9GB, 128K context

Name: Llama 3.1 Minitron 4B Width Base
Author: nvidia

Llama 3.1 Minitron 4B Width Base is an open-source language model by nvidia. Features: 4b LLM, VRAM: 9GB, Context: 128K, License: other, LLM Explorer Score: 0.18.

Arxiv:2009.03300 Arxiv:2407.14679 Arxiv:2408.11796 En Endpoints compatible Llama Llama-3 Nemo Nvidia Pytorch Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/nvidia/Llama-3.1-Minitron-4B-Width-Base

Llama 3.1 Minitron 4B Width Base Parameters and Internals

Model Type

text-to-text, generative

Use Cases

Areas:

research, commercial applications

Applications:

text generation, question answering

Primary Use Cases:

Paragraph completion

Limitations:

Potential for generating toxic, biased responses

Considerations:

Security measures and ethical reviews may be needed

Supported Languages

English (high), Multilingual (medium)

Training Details

Data Sources:

webpages, dialogue, articles

Data Volume:

94 billion tokens

Methodology:

Knowledge distillation

Context Length:

8000

Training Time:

July 29, 2024 and Aug 3, 2024

Hardware Used:

NVIDIA A100

Model Architecture:

Llama-3.1 core, Transformer Decoder

Responsible Ai Considerations

Transparency:

Some known biases

Accountability:

User responsibility

Mitigation Strategies:

Ensure requirements for use cases are met, unexpected misuse addressed

Input Output

Input Format:

String

Accepted Modalities:

Text

Output Format:

String

Performance Tips:

Use text prompts of 8000 characters or less for best results

Release Notes

Version:

Date:

2024-08-03

Notes:

Final deployment-ready version.

LLM Name	Llama 3.1 Minitron 4B Width Base
Repository 🤗	https://huggingface.co/nvidia/Llama-3.1-Minitron-4B-Width-Base
Model Size	4b
Required VRAM	9 GB
Updated	2026-07-17
Maintainer	nvidia
Model Type	llama
Model Files	5.0 GB: 1-of-2 4.0 GB: 2-of-2
Supported Languages	en
Model Architecture	LlamaForCausalLM
License	other
Context Length	131072
Model Max Length	131072
Transformers Version	4.45.0.dev0
Tokenizer Class	PreTrainedTokenizerFast
Vocabulary Size	128256
Torch Data Type	bfloat16

Rank the Llama 3.1 Minitron 4B Width Base capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Llama 3.1 Minitron 4B Width Base

Best Alternatives	Context / RAM	Downloads	Likes
4Bcpt	256K / 8.8 GB	5	0
HoldMy4BKTO	256K / 8.8 GB	5	0
Xgen Small 4B Instruct R	256K / 17.7 GB	54	4
Xgen Small 4B Base R	256K / 17.7 GB	30	3
SJT 4B	146K / 7.6 GB	7	0
...lama 3.1 Nemotron Nano 4B V1.1	128K / 9 GB	3940	116
Nemotron W 4b MagLight 0.1	128K / 9.2 GB	9	3
Loxa 4B	128K / 16 GB	8	0
Nemotron W 4b Halo 0.1	128K / 9.2 GB	22	3
Aura 4B	128K / 9 GB	44	14

Note: green Score (e.g. "73.2") means that the model is better than nvidia/Llama-3.1-Minitron-4B-Width-Base.