v2.1.1

2.2.0 (2024-10-14)

Features

• include training information README.md (edb150b)

Documentation

• update citation README.md (9454d91)

v2.1.0

2.1.0 (2024-08-14)

Features

• add confidence while prediction (fb8dc93)
• add detailed docstrings and documentation (9db9eef)
• add rdkit and readthedocs config (ca979c2)
• Add tests, improve workflow and add .gitignore (9255b41)
• Added support for conda package (2ae890b)
• create release-please.yml (cf2cb30)
• improved README.md (afd9dc3)
• Input SMILES Canonicalisation done using CDK instead of RDKit (c83e4f4)
• Remove image (ca0e2e2)
• updated models (d45654e)

Bug Fixes

• add missing requirements (cbe9519)
• conda install instructions (7ed9f0d)
• Do not show TensorFlow warnings while loading (ecc8687)
• docs: fixed typo in readMe (945478e)
• Error with documentation imports (5b4675a)
• GitHub url (56cb302)
• import error tensorflow (149cd30)
• include doctree information (b58e56a)
• Installation instructions (ba5daf4)
• model input length (61035c5)
• Remove RDKit requirement (220a93c)
• Remove subprocess unzip (e7a5eca)
• remove unused imports (84a6e06)
• Remove unused imports (13ae6b8)
• Remove unwanted requirements (1eab987)
• Run tests only with Python 3.10 (68cd77e)
• support for python versions 3.7 and above (e7bb840)
• Update DOI (6dec565)
• update logo README.md (3fa5a41)
• update requirements.txt (105ea95)
• Update TRC logo (518537b)
• updated docstrings (7056acc)
• use rdkit, improve fucntions and resolve issues (786fe32)

STOUT V2.0.3

• Remove RDKit imports
• Minor bug fixes
• Remove ubuntu unzip dependency and use python default zipfile

Full Changelog: V2.0.2...V2.0.3

STOUT V2.0.2

What's Changed

• Canonicalisation on input SMILES done using CDK
• Newly deployed model in the backend
• Maximum number of input and output tokens length increased
• Tests using pytest introduced
• Models now trained using TPU V4-VMs

Full Changelog: V2.0.0...V2.0.2

STOUT-V2 published

STOUT V2

• Smiles TO iUpac Translator now uses Transformers at the backend
• Trained on PubChem IUPAC names
• Model optimized to generate names fast
• Translation performance is better than the previous version
• The models were Trained on TPU v3-8 nodes
• Packaged to install directly from Pypi
• Docstrings added
• web version is available at https://decimer.ai

Initial release to archive in Zenodo

V2.1.1

Update README.md

Initial Release

Version 2.1 updates:
IUPAC names were split into tokens using an advanced ruleset.
Newly trained models updated.
Tokenizer modified to fit the advanced ruleset.
selfies decoding updated, 1.0.4 version included.