SMILES Tokenizer

This is a custom tokenizer for SMILES (Simplified Molecular Input Line Entry System) strings.

Usage

from transformers import AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained('suku9/smiles-tokenizer')

# Tokenize a SMILES string
tokens = tokenizer.tokenize('CC(=O)OC1=CC=CC=C1C(=O)O')
print(tokens)

# Encode a SMILES string
encoded = tokenizer('CC(=O)OC1=CC=CC=C1C(=O)O', return_tensors='pt')
print(encoded)

# Use with a GPT-2 model
from transformers import GPT2LMHeadModel

model = GPT2LMHeadModel.from_pretrained('gpt2')
# Resize model embeddings to match tokenizer vocabulary size
model.resize_token_embeddings(len(tokenizer))

Features

Specialized for SMILES strings
Compatible with Hugging Face's transformers library
Designed to work with GPT-2 models
Preserves all functionality of the original SMILES tokenizer