What are the hardware requirements for NuExtract Tiny?

NuExtract Tiny requires approximately 1.9 GB of VRAM and supports a context window of 32K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed NuExtract Tiny and how large is it?

NuExtract Tiny is developed by numind, a model with 0.5b 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 NuExtract Tiny?

NuExtract Tiny is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

NuExtract Tiny by numind — VRAM 1.9GB, 32K context

Name: NuExtract Tiny
Author: numind

NuExtract Tiny is an open-source language model by numind. Features: 0.5b LLM, VRAM: 1.9GB, Context: 32K, License: mit, LLM Explorer Score: 0.17.

Base model:finetune:qwen/qwen1... Base model:qwen/qwen1.5-0.5b Conversational Deploy:azure En Endpoints compatible Qwen2 Region:us Safetensors

Model Card on HF 🤗: https://huggingface.co/numind/NuExtract-tiny

NuExtract Tiny Parameters and Internals

Model Type

information extraction

Use Cases

Limitations:

Model is outdated

Additional Notes

The model performs well in zero-shot settings but is intended for fine-tuning on specific tasks.

Supported Languages

en (Supported)

Training Details

Data Sources:

high-quality synthetic dataset

Methodology:

fine-tuning

Context Length:

2000

Hardware Used:

GPU

Input Output

Input Format:

Text (less than 2000 tokens) and a JSON template

Accepted Modalities:

text

Output Format:

Extracted information based on template

Performance Tips:

Fine-tune on a specific task with 30+ examples for better performance

LLM Name	NuExtract Tiny
Repository 🤗	https://huggingface.co/numind/NuExtract-tiny
Base Model(s)	Qwen/Qwen1.5-0.5B Qwen/Qwen1.5-0.5B
Model Size	0.5b
Required VRAM	1.9 GB
Updated	2026-08-08
Maintainer	numind
Model Type	qwen2
Model Files	1.9 GB
Supported Languages	en
Model Architecture	Qwen2ForCausalLM
License	mit
Context Length	32768
Model Max Length	32768
Transformers Version	4.42.0.dev0
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Vocabulary Size	151936
Torch Data Type	float32
Errors	replace

Rank the NuExtract Tiny 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 NuExtract Tiny

Best Alternatives	Context / RAM	Downloads	Likes
Reader Lm 0.5B	250K / 1 GB	188	150
Qwen2 0.5B	128K / 1 GB	1216780	169
Qwen2 0.5B Finetune Exp 1	128K / 1 GB	15	0
...0.5B FT EnhancedMUSREnsembleV3	128K / 1 GB	32	1
....5B FT V4 MUSR ENSEMBLE Mathis	128K / 2 GB	30	1
...0.5B FT MUSR ENSEMBLE V2Mathis	128K / 1 GB	25	0
...PRYMMAL 0.5B FT V4 MUSR Mathis	128K / 1 GB	62	1
Qwen2 0.5B Stf Imdb	128K / 1 GB	7	0
ECE PRYMMAL 0.5B FT V3 MUSR	128K / 2 GB	17	0
ECE PRYMMAL 0.5B FT V4 MUSR	128K / 2 GB	10	0

Note: green Score (e.g. "73.2") means that the model is better than numind/NuExtract-tiny.