Skip to content
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.

Sign up for GitHub

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

Add explicit batch size parameter for FID computation #110

Merged
merged 1 commit into from
Mar 2, 2024
Merged

Add explicit batch size parameter for FID computation #110

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 4 additions & 2 deletions examples/images/cifar10/compute_fid.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,8 @@
flags.DEFINE_integer("step", 400000, help="training steps")
flags.DEFINE_integer("num_gen", 50000, help="number of samples to generate")
flags.DEFINE_float("tol", 1e-5, help="Integrator tolerance (absolute and relative)")
flags.DEFINE_float("batch_size_fid", 1024, help="Batch size to compute FID")

FLAGS(sys.argv)


Expand Down Expand Up @@ -70,7 +72,7 @@

def gen_1_img(unused_latent):
with torch.no_grad():
x = torch.randn(500, 3, 32, 32, device=device)
x = torch.randn(FLAGS.batch_size_fid, 3, 32, 32, device=device)
if FLAGS.integration_method == "euler":
print("Use method: ", FLAGS.integration_method)
t_span = torch.linspace(0, 1, FLAGS.integration_steps + 1, device=device)
Expand All @@ -90,7 +92,7 @@ def gen_1_img(unused_latent):
score = fid.compute_fid(
gen=gen_1_img,
dataset_name="cifar10",
batch_size=500,
batch_size=FLAGS.batch_size_fid,
dataset_res=32,
num_gen=FLAGS.num_gen,
dataset_split="train",
Expand Down
Loading