Skip to content

Partially unfreeze MERL text encoder #25

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

Open
TonyChen06 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Partially unfreeze MERL text encoder #25

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
9 changes: 5 additions & 4 deletions src/neural_networks/merl/merl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,7 @@ class MerlConfig:
dropout: float = 0.1
seq_len: int = 2500
lm: str = "ncbi/MedCPT-Query-Encoder"
frozen_text_layers: int = 6
resnet_type: str = "resnet101"
distributed: bool = False
spacial_dim: int = None
Expand All @@ -42,8 +43,9 @@ def __init__(self, cfg: MerlConfig):
self.cfg = cfg
self.resnet = get_resnet(cfg.resnet_type)
self.lm = AutoModel.from_pretrained(cfg.lm)
for p in self.lm.parameters():
p.requires_grad = False
for layer in self.lm.encoder.layer[:cfg.frozen_text_layers]:
for p in layer.parameters():
p.requires_grad = False

self.downconv = nn.Conv1d(cfg.in_channels, cfg.proj_out, kernel_size=1)
self.att_pool_head = AttentionPool2d(
Expand Down Expand Up @@ -72,8 +74,7 @@ def forward(self, signal: torch.Tensor, condition: dict):
ecg1 = self.dropout1(self.linear1(ecg_pooled))
ecg2 = self.dropout2(self.linear2(ecg_pooled))

with torch.no_grad():
text_emb = self.lm(**condition).pooler_output
text_emb = self.lm(**condition).pooler_output
proj_text = self.proj_t(text_emb)

if self.cfg.distributed:
Expand Down