What are the hardware requirements for Csmpt7b?

Csmpt7b requires approximately 26.7 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Csmpt7b and how large is it?

Csmpt7b is developed by BUT-FIT, a model with 6.7b 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 Csmpt7b?

Csmpt7b is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Csmpt7b by BUT-FIT — VRAM 26.7GB

Name: Csmpt7b
Author: BUT-FIT

Csmpt7b is an open-source language model by BUT-FIT. Features: 6.7b LLM, VRAM: 26.7GB, License: apache-2.0, Quantized, LLM Explorer Score: 0.12.

Arxiv:2304.09871 Arxiv:2412.17933 Cs Custom code Dataset:but-fit/adult content ... Dataset:but-fit/but-lcc Endpoints compatible Gguf Gguf-my-repo Llama-cpp Mpt Q8 Quantized Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/BUT-FIT/csmpt7b

Csmpt7b Parameters and Internals

Model Type

text generation

Use Cases

Areas:

Czech language processing

Applications:

academic research, commercial applications

Primary Use Cases:

text generation for Czech language content

Additional Notes

Vocabulary swap was utilized to enhance transfer of knowledge from English to Czech.

Supported Languages

Czech (native)

Training Details

Data Sources:

BUT-FIT/BUT-LCC, BUT-FIT/adult_content_classifier_dataset

Data Volume:

272 billion training tokens, with ~67 billion tokens specific to Czech

Methodology:

Vocabulary swap method aligning English tokens to Czech equivalents

Context Length:

2048

Hardware Used:

Karolina cluster

Model Architecture:

Continuation of MPT7b with adaptations for Czech language

Input Output

Input Format:

Czech textual prompts

Accepted Modalities:

text

Output Format:

Generated Czech text

Release Notes

Version:

N/A

Date:

18/04/2024

Notes:

Release of training checkpoints

Version:

N/A

Date:

06/05/2024

Notes:

Release of manually annotated dataset for adult content classification

LLM Name	Csmpt7b
Repository 🤗	https://huggingface.co/BUT-FIT/csmpt7b
Model Size	6.7b
Required VRAM	26.7 GB
Updated	2026-09-17
Maintainer	BUT-FIT
Model Type	mpt
Model Files	13.4 GB 26.8 GB 7.1 GB 4.8 GB: 1-of-6 4.8 GB: 2-of-6 4.8 GB: 3-of-6 4.8 GB: 4-of-6 4.8 GB: 5-of-6 2.7 GB: 6-of-6
Supported Languages	cs
GGUF Quantization	Yes
Quantization Type	gguf\|q8
Model Architecture	MPTForCausalLM
License	apache-2.0
Model Max Length	2048
Transformers Version	4.37.0.dev0
Tokenizer Class	PreTrainedTokenizerFast
Padding Token	[EOS]
Vocabulary Size	64002
Torch Data Type	float32

Rank the Csmpt7b 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