search

gligen / GLIGEN

Open-Set Grounded Text-to-Image Generation
MIT License
1.91k stars 145 forks source link

Is it possible to train GLIGEN with a single GPU? #48

Closed YJ-142150 closed 10 months ago

YJ-142150 commented 11 months ago

Thanks for your great work! Is it possible to train GLIGEN with 3090TI?

crj1998 commented 10 months ago

yes, with fp16 it vram usage is smaller than 24GB

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.57.02    Driver Version: 470.57.02    CUDA Version: 11.7     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  NVIDIA A100-PCI...  Off  | 00000000:8B:00.0 Off |                    0 |
| N/A   49C    P0    96W / 250W |  16052MiB / 40536MiB |    100%      Default |
|                               |                      |             Disabled |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
+-----------------------------------------------------------------------------+
YJ-142150 commented 10 months ago

Thank you for answering the question!!

Sjey-Lyn commented 6 months ago

How long the training takes with 3090TI?

yejinxueshi commented 5 months ago

是的，对于 fp16，vram 使用量小于 24GB

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.57.02    Driver Version: 470.57.02    CUDA Version: 11.7     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  NVIDIA A100-PCI...  Off  | 00000000:8B:00.0 Off |                    0 |
| N/A   49C    P0    96W / 250W |  16052MiB / 40536MiB |    100%      Default |
|                               |                      |             Disabled |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
+-----------------------------------------------------------------------------+

Where should I change the accuracy to fp16