shen-lab / ldm-3dg Goto Github PK

Hi Authors,

Congrats on the great work!

As I was reading your team's paper about 3DG, I want to reproduce your excellent work. However, I noticed that you didn't specify the GPU version used to train your model in the paper. Additionally, I observed that the default batch size is 512 and the number of epochs is 10,000. These numbers seem larger than those used in other models I have encountered before. I couldn't find the exact numbers you used to train the unconditional QM9 model in the paper. Therefore, I am wondering: What GPU version did you use to train the unconditional QM9 model, and are the batch size and number of epochs really 512 and 10,000?

Thanks!

Hi Authors,

Congrats on the great work! I am trying to develop a flow-matching version of your work, but I encounter some challenges when reproducing your work on Conditional Generation on Geometric Objects.

When I ran Sampling and evaluating commands, I encountered the following error for all data_id I tried.

[2024-05-08 23:53:49,959::evaluate::INFO] Vina Score:  Mean: nan Median: nan
[2024-05-08 23:53:49,959::evaluate::INFO] Vina Min  :  Mean: nan Median: nan
Traceback (most recent call last):
  File "/global/cfs/cdirs/mp54/jsliang/MLFF/conda_envs/eval/lib/python3.9/runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/global/cfs/cdirs/mp54/jsliang/MLFF/conda_envs/eval/lib/python3.9/runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "/global/cfs/cdirs/mp54/jsliang/LFM-3DG/AE_Geometry_and_Conditional_Latent_Diffusion/scripts/evaluate.py", line 154, in <module>
    print_ring_ratio([r['chem_results']['ring_size'] for r in results], logger)
  File "/global/cfs/cdirs/mp54/jsliang/LFM-3DG/AE_Geometry_and_Conditional_Latent_Diffusion/scripts/evaluate.py", line 34, in print_ring_ratio
    logger.info(f'ring size: {ring_size} ratio: {n_mol / len(all_ring_sizes):.3f}')
ZeroDivisionError: division by zero

What I changed are:

1. the model used in sample_z because I do not have the folder ldm_2023_11_16__18_01_30
   model.load_state_dict(torch.load('logs_diffusion/ldm_2023_11_16__18_01_30/checkpoints/30000.pt')['model'])
   to
   model.load_state_dict(torch.load('../AE_geom_cond_weights_and_data/weight_diffusion.pt')['model'])
2. the version name from 'final-001' to 'final' in pl_pair_dataset.py to match the file name in https://github.com/guanjq/targetdiff#data
   def __init__(self, raw_path, transform=None, version='final'):
   Can you help check what's wrong?

Hi Yuning,

I am very interesting in your great work, and I'm attempting to reproduce your results using the Jupyter notebook. However, I've encountered some challenges with the 2D distribution results while using the provided sampled SMILES data.

Specifically, I directly evaluated AE_geom_uncond_weights_and_data/job17_latent_ddpm_qm9_spatial_graphs/sample_smiles.pt and sample_conformer.pt. For the test SMILES, I loaded e3_diffusion_for_molecules/data/smiles_qm9.txt. All the files were sourced from your comprehensive Zenodo repository.

I've attached a figure showing the results I obtained. Could you kindly advise if I'm using the correct data or provide any guidance on the appropriate files to use for reproducing the 2D distribution results?

I got this error when trying to unzip the file:

error [AE_geom_cond_weights_and_data.zip]: start of central directory not found;
zipfile corrupt.