ASTRA: Optimized SMILES Generator (Final)

This model is part of the ASTRA (Advanced Solvation Transformer for Rational Additives) framework. It is a GPT-2 based Causal Language Model (CausalLM) that has been fully optimized via an Active Learning loop. Unlike the initial generator, this final model has been fine-tuned to strategically generate novel, high-reward electrolyte additive molecules that target specific Lowest Unoccupied Molecular Orbital (LUMO) and Binding Energy (Eb) values.

Model Details

• Architecture: GPT-2 (CausalLM)
• Stage: Final (After Active Learning Loop)
• Task: Targeted Autoregressive SMILES Generation
• Optimization: Active Learning with property-based reward signals

Usage

from astra.tokenizer import initial_gpt_tokenizer_with_vocabulary
from astra.model import GPTForCausalLM

# Load the tokenizer and model
vocab_file = "./vocab.txt"
tokenizer = initial_gpt_tokenizer_with_vocabulary(path=vocab_file)

pretrained_path = "<path/to/pretrained/model>" # Replace with your actual paths
model = GPTForCausalLM.from_pretrained(pretrained_gpt_path)

More Information

For more details on data preparation, downstream fine-tuning, and the full active learning loop, please visit our GitHub Repository.