European Open Source AI Index

Database News Guides About Contribute

CodeT5+

by Salesforce

About the model:

T5-based open code model. Model components initialized with CodeGen.

Model type:

Code

Model performance class:

Full

Link to the model:

https://huggingface.co/Salesforce/instructcodet5p-16b

Base models:

CodeGen-16B-NL

End model:

InstructCodeT5p-16B

End model license:

BSD 3-Clause

About the organisation:

Major AI research company.

Link to the organisation:

https://www.salesforceairesearch.com/

Model release date:

May 2023

Availability

Base Model Data

Model trained on The Pile, BigQuery datasets, and BigQuery. Not all datasets were found.

https://pile.eleuther.ai/https://cloud.google.com/bigquery/docs/datasets-intro

End User Model Data

Pretraining data from CodeSearchNet and GitHub Code. Instruction-tuning data from CodeAlpaca.

https://huggingface.co/datasets/code-search-net/code_search_net https://huggingface.co/datasets/codeparrot/github-code https://github.com/sahil280114/codealpaca

Base Model Weights

Model weights made available on HuggingFace.

https://huggingface.co/Salesforce/codegen-16B-nl

End User Model Weights

Model weights made available on HuggingFace.

https://huggingface.co/Salesforce/instructcodet5p-16b

Training Code

Repo exists, but does not contain training code.

https://github.com/salesforce/CodeT5/tree/main

Documentation

Code Documentation

No training code, so no documentation.

Hardware Architecture

Hardware architecture described in preprint.

https://arxiv.org/pdf/2305.07922

Preprint

Preprint available on arXiv.

https://arxiv.org/pdf/2305.07922

Paper

Paper published as conference paper.

https://aclanthology.org/2023.emnlp-main.68/

Modelcard

Model card provides the requisite detail.

https://huggingface.co/Salesforce/instructcodet5p-16b

Datasheet

Data sheets available for most data sources, with varying levels of detail regarding collection and curation procedures.

https://pile.eleuther.ai/https://cloud.google.com/bigquery/docs/datasets-intro https://huggingface.co/datasets/code-search-net/code_search_net https://huggingface.co/datasets/codeparrot/github-code https://github.com/sahil280114/codealpaca

Access

Licenses

BSD 3-Clause, an open-source license.

https://huggingface.co/Salesforce/instructcodet5p-16b

Is this information not up to date?

Contribute here ->

Supported by the Centre for Language Studies and the Dutch Research Council. Website design & development © 2024 by BSTN. This version of the index generated 09 April 2026, website content last updated 11 March 2026.