Skip to content

[Tests] reduce the model size in the amused fast test #7804

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

Merged
sayakpaul merged 7 commits into from
Apr 30, 2024
Merged

[Tests] reduce the model size in the amused fast test #7804

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
fff6598
chore: reducing model sizes
ariG23498 Apr 28, 2024
136577d
chore: shrinks further
ariG23498 Apr 29, 2024
ab4339c
chore: shrinks further
ariG23498 Apr 29, 2024
407009a
chore: shrinking model for img2img pipeline
ariG23498 Apr 29, 2024
2eb50fe
chore: reducing size of model for inpaint pipeline
ariG23498 Apr 29, 2024
e7458b9
Merge branch 'main' into aritra/amused-test-fast
sayakpaul Apr 29, 2024
f86cf1f
Merge branch 'main' into aritra/amused-test-fast
sayakpaul Apr 30, 2024
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
36 changes: 18 additions & 18 deletions tests/pipelines/amused/test_amused.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,43 +38,43 @@ class AmusedPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
def get_dummy_components(self):
torch.manual_seed(0)
transformer = UVit2DModel(
hidden_size=32,
hidden_size=8,
use_bias=False,
hidden_dropout=0.0,
cond_embed_dim=32,
cond_embed_dim=8,
micro_cond_encode_dim=2,
micro_cond_embed_dim=10,
encoder_hidden_size=32,
encoder_hidden_size=8,
vocab_size=32,
codebook_size=32,
in_channels=32,
block_out_channels=32,
codebook_size=8,
in_channels=8,
block_out_channels=8,
num_res_blocks=1,
downsample=True,
upsample=True,
block_num_heads=1,
num_hidden_layers=1,
num_attention_heads=1,
attention_dropout=0.0,
intermediate_size=32,
intermediate_size=8,
layer_norm_eps=1e-06,
ln_elementwise_affine=True,
)
scheduler = AmusedScheduler(mask_token_id=31)
torch.manual_seed(0)
vqvae = VQModel(
act_fn="silu",
block_out_channels=[32],
block_out_channels=[8],
down_block_types=[
"DownEncoderBlock2D",
],
in_channels=3,
latent_channels=32,
layers_per_block=2,
norm_num_groups=32,
num_vq_embeddings=32,
latent_channels=8,
layers_per_block=1,
norm_num_groups=8,
num_vq_embeddings=8,
out_channels=3,
sample_size=32,
sample_size=8,
up_block_types=[
"UpDecoderBlock2D",
],
Expand All @@ -85,14 +85,14 @@ def get_dummy_components(self):
text_encoder_config = CLIPTextConfig(
bos_token_id=0,
eos_token_id=2,
hidden_size=32,
intermediate_size=64,
hidden_size=8,
intermediate_size=8,
layer_norm_eps=1e-05,
num_attention_heads=8,
num_hidden_layers=3,
num_attention_heads=1,
num_hidden_layers=1,
pad_token_id=1,
vocab_size=1000,
projection_dim=32,
projection_dim=8,
)
text_encoder = CLIPTextModelWithProjection(text_encoder_config)
tokenizer = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip")
Expand Down
36 changes: 18 additions & 18 deletions tests/pipelines/amused/test_amused_img2img.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,43 +42,43 @@ class AmusedImg2ImgPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
def get_dummy_components(self):
torch.manual_seed(0)
transformer = UVit2DModel(
hidden_size=32,
hidden_size=8,
use_bias=False,
hidden_dropout=0.0,
cond_embed_dim=32,
cond_embed_dim=8,
micro_cond_encode_dim=2,
micro_cond_embed_dim=10,
encoder_hidden_size=32,
encoder_hidden_size=8,
vocab_size=32,
codebook_size=32,
in_channels=32,
block_out_channels=32,
codebook_size=8,
in_channels=8,
block_out_channels=8,
num_res_blocks=1,
downsample=True,
upsample=True,
block_num_heads=1,
num_hidden_layers=1,
num_attention_heads=1,
attention_dropout=0.0,
intermediate_size=32,
intermediate_size=8,
layer_norm_eps=1e-06,
ln_elementwise_affine=True,
)
scheduler = AmusedScheduler(mask_token_id=31)
torch.manual_seed(0)
vqvae = VQModel(
act_fn="silu",
block_out_channels=[32],
block_out_channels=[8],
down_block_types=[
"DownEncoderBlock2D",
],
in_channels=3,
latent_channels=32,
layers_per_block=2,
norm_num_groups=32,
num_vq_embeddings=32,
latent_channels=8,
layers_per_block=1,
norm_num_groups=8,
num_vq_embeddings=32, # reducing this to 16 or 8 -> RuntimeError: "cdist_cuda" not implemented for 'Half'
out_channels=3,
sample_size=32,
sample_size=8,
up_block_types=[
"UpDecoderBlock2D",
],
Expand All @@ -89,14 +89,14 @@ def get_dummy_components(self):
text_encoder_config = CLIPTextConfig(
bos_token_id=0,
eos_token_id=2,
hidden_size=32,
intermediate_size=64,
hidden_size=8,
intermediate_size=8,
layer_norm_eps=1e-05,
num_attention_heads=8,
num_hidden_layers=3,
num_attention_heads=1,
num_hidden_layers=1,
pad_token_id=1,
vocab_size=1000,
projection_dim=32,
projection_dim=8,
)
text_encoder = CLIPTextModelWithProjection(text_encoder_config)
tokenizer = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip")
Expand Down
36 changes: 18 additions & 18 deletions tests/pipelines/amused/test_amused_inpaint.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,43 +42,43 @@ class AmusedInpaintPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
def get_dummy_components(self):
torch.manual_seed(0)
transformer = UVit2DModel(
hidden_size=32,
hidden_size=8,
use_bias=False,
hidden_dropout=0.0,
cond_embed_dim=32,
cond_embed_dim=8,
micro_cond_encode_dim=2,
micro_cond_embed_dim=10,
encoder_hidden_size=32,
encoder_hidden_size=8,
vocab_size=32,
codebook_size=32,
in_channels=32,
block_out_channels=32,
codebook_size=32, # codebook size needs to be consistent with num_vq_embeddings for inpaint tests
in_channels=8,
block_out_channels=8,
num_res_blocks=1,
downsample=True,
upsample=True,
block_num_heads=1,
num_hidden_layers=1,
num_attention_heads=1,
attention_dropout=0.0,
intermediate_size=32,
intermediate_size=8,
layer_norm_eps=1e-06,
ln_elementwise_affine=True,
)
scheduler = AmusedScheduler(mask_token_id=31)
torch.manual_seed(0)
vqvae = VQModel(
act_fn="silu",
block_out_channels=[32],
block_out_channels=[8],
down_block_types=[
"DownEncoderBlock2D",
],
in_channels=3,
latent_channels=32,
layers_per_block=2,
norm_num_groups=32,
num_vq_embeddings=32,
latent_channels=8,
layers_per_block=1,
norm_num_groups=8,
num_vq_embeddings=32, # reducing this to 16 or 8 -> RuntimeError: "cdist_cuda" not implemented for 'Half'
out_channels=3,
sample_size=32,
sample_size=8,
up_block_types=[
"UpDecoderBlock2D",
],
Expand All @@ -89,14 +89,14 @@ def get_dummy_components(self):
text_encoder_config = CLIPTextConfig(
bos_token_id=0,
eos_token_id=2,
hidden_size=32,
intermediate_size=64,
hidden_size=8,
intermediate_size=8,
layer_norm_eps=1e-05,
num_attention_heads=8,
num_hidden_layers=3,
num_attention_heads=1,
num_hidden_layers=1,
pad_token_id=1,
vocab_size=1000,
projection_dim=32,
projection_dim=8,
)
text_encoder = CLIPTextModelWithProjection(text_encoder_config)
tokenizer = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip")
Expand Down