PhTae/MolBridge-Gen-Base-C2S

EMNLP 2025 main

"Bridging the Gap Between Molecule and Textual Descriptions via Substructure-aware Alignment"

GitHub Paper

This model is trained on ChEBI-20 dataset.

from transformers import AutoTokenizer, T5ForConditionalGeneration
tokenizer = AutoTokenizer.from_pretrained("laituan245/molt5-base", model_max_length=512)
model = T5ForConditionalGeneration.from_pretrained('PhTae/MolBridge-Gen-Base-C2S')

caption = 'The molecule is a monoterpene that is bicyclo[2.2.1]heptane substituted by methy groups at positions 1, 3 and 3. It is a monoterpene, a terpenoidfundamental parent and a carbobicyclic compound.'
caption = 'Provide a molecule based on this description: ' + caption

token = tokenizer(caption, return_tensors='pt', padding='longest', truncation=True)
gen_results = model.generate(input_ids=token['input_ids'],
                             attention_mask=token['attention_mask'], 
                             num_beams=5, 
                             max_new_tokens=512)

gen_results = tokenizer.decode(gen_results[0], skip_special_tokens=True)
print(gen_results)