-
Notifications
You must be signed in to change notification settings - Fork 113
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
model collapse and image blurring #88
Comments
I had this same issue and its caused either by : if your scores/fake = NaN that means you have a library/plugin/mkl conflict |
@mapengsen please share the text of your stats.jsonl file. |
{"Loss/scores/fake": {"num": 1312000, "mean": -4.672473785028225, "std": 2.1516482762262013}, "Loss/signs/fake": {"num": 1312000, "mean": -0.9828719512195122, "std": 0.18428979219139854}, "Loss/G/loss": {"num": 2000, "mean": 18.6724349565506, "std": 3.7821310307624945}, "Loss/scores/real": {"num": 656000, "mean": 5.545181464427855, "std": 2.436094622720647}, "Loss/signs/real": {"num": 656000, "mean": 0.9719756097560975, "std": 0.23508171779035134}, "Loss/D/loss": {"num": 2000, "mean": 0.2549195118771688, "std": 0.5003099562855534}, "Progress/tick": {"num": 1, "mean": 40.0, "std": 0.0}, "Progress/kimg": {"num": 1, "mean": 164.0019989013672, "std": 0.0}, "Timing/total_sec": {"num": 1, "mean": 1702.18701171875, "std": 0.0}, "Timing/sec_per_tick": {"num": 1, "mean": 1635.7618408203125, "std": 0.0}, "Timing/sec_per_kimg": {"num": 1, "mean": 408.9404602050781, "std": 0.0}, "Timing/maintenance_sec": {"num": 1, "mean": 66.42518615722656, "std": 0.0}, "Resources/cpu_mem_gb": {"num": 1, "mean": 5.329277038574219, "std": 0.0}, "Resources/peak_gpu_mem_gb": {"num": 1, "mean": 9.622943878173828, "std": 0.0}, "Resources/peak_gpu_mem_reserved_gb": {"num": 1, "mean": 9.755859375, "std": 0.0}, "Progress/augment": {"num": 1, "mean": 0.0, "std": 0.0}, "Timing/total_hours": {"num": 1, "mean": 0.4728297293186188, "std": 0.0}, "Timing/total_days": {"num": 1, "mean": 0.019701238721609116, "std": 0.0}, "Timing/Gmain": {"num": 1, "mean": 550.1205444335938, "std": 0.0}, "Timing/Greg": {"num": 1, "mean": 0.001728000002913177, "std": 0.0}, "Timing/Dmain": {"num": 1, "mean": 257.69061279296875, "std": 0.0}, "Timing/Dreg": {"num": 1, "mean": 4.30079984664917, "std": 0.0}, "timestamp": 1662131910.3168495} |
I want to generate molecular images based on molecular images, but after I trained 80kimg, the model began to collapse, and the generated images were not clear. How many kimgs did you train at that time and then began to converge?
real image:
generate image :
The text was updated successfully, but these errors were encountered: