Skip to content

a few more fixes #3

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
will-rice merged 1 commit into from
Jan 8, 2026
Merged

a few more fixes #3

Show file tree
Hide file tree
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: 3 additions & 3 deletions src/syncnet/config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ class Config(BaseModel):

Attributes:
seed: Random seed for reproducibility across runs.
test_split: Fraction of the dataset to use for validation/testing.
val_split: Fraction of the dataset to use for validation/testing.
batch_size: Number of samples per batch during training.
max_epochs: Maximum number of training epochs.
early_stopping_patience: Number of epochs to wait before early stopping.
Expand All @@ -43,8 +43,8 @@ class Config(BaseModel):
seed: int = Field(default=42, description="Random seed for reproducibility.")

# Data
test_split: float = Field(
default=0.05, description="Proportion of data for val/testing."
val_split: float = Field(
default=0.05, description="Proportion of data for validation."
)
batch_size: int = Field(default=4, description="Batch size.")

Expand Down
15 changes: 9 additions & 6 deletions src/syncnet/datamodule.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -62,7 +62,10 @@ def __init__(self, dataset: Dataset, config: Config, num_workers: int = 4) -> No
self.config = config
self.num_workers = num_workers
self.processor = PeAudioVideoProcessor.from_pretrained(config.base_model)
self.resample = torchaudio.transforms.Resample(16000, 48000)
self.resample = torchaudio.transforms.Resample(
16000,
self.processor.feature_extractor.sampling_rate, # type: ignore[attr-defined]
)
self.resize = torchvision.transforms.Resize(
(config.frame_height, config.frame_width)
)
Expand All @@ -79,7 +82,7 @@ def setup(self, stage: str | None = None) -> None:
"""
if stage == "fit" or stage is None:
dataset_size = len(self.dataset) # type: ignore
train_size = int(0.8 * dataset_size)
train_size = int((1.0 - self.config.val_split) * dataset_size)
val_size = dataset_size - train_size
self.train_dataset, self.val_dataset = torch.utils.data.random_split(
self.dataset, [train_size, val_size]
Expand Down Expand Up @@ -156,13 +159,13 @@ def pad_collate_fn(
audio = audio.mean(dim=0, keepdim=True) # Convert to mono if stereo

# Randomly create negative samples by mixing audio/video
mix = random.random() > 0.5
mix = random.random() > self.config.negative_fraction
video_segment, audio_segment = self.sample_random_segment(
video,
audio,
num_frames=self.config.num_frames,
fps=25,
sample_rate=16000,
fps=metadata.get("video_fps", 25),
sample_rate=metadata.get("audio_fps", 16000),
mix=mix,
)
if audio_segment.shape[1] < 3200:
Expand All @@ -177,7 +180,7 @@ def pad_collate_fn(
audio=audio_segment.squeeze(0),
return_tensors="pt",
padding=False,
sampling_rate=48000,
sampling_rate=self.processor.feature_extractor.sampling_rate, # type: ignore[attr-defined]
)
video_segments.append(input_values["pixel_values_videos"][0])
audio_segments.append(input_values["input_values"][0])
Expand Down