Awesome

DiffSeqMol

DiffSeqMol: Diffusion model for molecular sequence generation

Unconditional Genration

If you want to use optimizition code, please switch to the master branch

train

sample

Optimizition

If you want to use optimizition code, please switch to the opt branch

train

CUDA_VISIBLE_DEVICES=0  python -m torch.distributed.launch --nproc_per_node=1 --master_port=port
  --use_env run_train.py 
  --diff_steps diff_steps 
  --lr learing_rate 
  --learning_steps learning_step
  --save_interval save_interval 
  --seed 102 
  --noise_schedule sqrt 
  --hidden_dim hidden_dim 
  --bsz batch_size 
  --dataset data 
  --data_dir work_dir  
  --vocab bert 
  --seq_len seq_len 
  --schedule_sampler lossaware 
  --notes qqp 
  --config_name seyonec/PubChem10M_SMILES_BPE_450k  
  --microbatch batch_size

sample

CUDA_VISIBLE_DEVICES=0 python -m torch.distributed.launch --nproc_per_node=1 --master_port=12356 
--use_env sample_seq2seq.py 
--model_path  model_path
--step diff_steps
--batch_size batch_size 
--seed2 123 
--split valid 
--out_dir generation_outputs 
--top_p -1