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

Update model_dd.py #1

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Update model_dd.py #1

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
635dd7e
Update model_dd.py
Maxwe11y Nov 30, 2022
2a37563
Update model_dd.py
Maxwe11y Dec 1, 2022
f8ae1ff
Merge branch 'main' into patch-1
Maxwe11y Dec 1, 2022
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
58 changes: 4 additions & 54 deletions model_dd.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -59,17 +59,10 @@ def __init__(self, cpt_ids, Configs, cuda_=True):

if self.use_future:
self.relAtt = RelAtt(1, 1, (self.window, self.input_dim), heads=self.num_head, dim_head=self.input_dim, dropout=Configs.att_dropout)
# self.relAtt = RelAtt(self.window, 1, (1, self.input_dim), heads=self.num_head, dim_head=self.input_dim,
# dropout=Configs.att_dropout)

else:
self.relAtt = RelAtt(1, 1, (self.slide_win+1, self.input_dim), heads=self.num_head, dim_head=self.input_dim,
dropout=Configs.att_dropout)
# self.relAtt = RelAtt(self.slide_win+1, 1, (1, self.input_dim), heads=self.num_head, dim_head=self.input_dim,
# dropout=Configs.att_dropout)


# self.relAtt = Trans_RelAtt(1, 1, (self.window, self.input_dim), heads=self.num_head, dim_head=self.input_dim // 2,
# dropout=Configs.att_dropout)

self.r = nn.Parameter(nn.init.uniform_(torch.zeros(3, self.input_dim)), requires_grad=True)
self.num_feature = Configs.num_features
Expand Down Expand Up @@ -121,9 +114,7 @@ def __init__(self, cpt_ids, Configs, cuda_=True):
self.input_dim)) # nn.ParameterList([nn.Parameter(torch.randn(self.input_dim, self.input_dim)) for _ in range(3)])

def forward(self, inputs, str_src, str_dst, str_edge_type, chunks, label, loss_func, train=True, eps=1e-8):
# torch.autograd.set_detect_anomaly(True)

# len_dial = len(inputs['input_ids'])

if self.model_type == 'albert':
out = self.bert_encoder(input_ids=inputs['input_ids'], attention_mask=inputs['attention_mask'],
token_type_ids=inputs['token_type_ids'])
Expand Down Expand Up @@ -196,8 +187,6 @@ def forward(self, inputs, str_src, str_dst, str_edge_type, chunks, label, loss_f

src_mask = torch.sum(masks, dim=-1) > 0
att_score = torch.softmax(self.get_att_masked(dot_sum, src_mask), dim=-1) * src_masks.ne(0)
# att_score = torch.softmax(dot_sum, dim=-1) * src_mask
# sent_mask_sum = torch.sum(src_masks.sum(dim=-1).ne(0)) + eps
symbolic_repr = torch.sum(att_score.unsqueeze(2) * src_emb, dim=1) # /sent_mask_sum


Expand All @@ -223,52 +212,13 @@ def forward(self, inputs, str_src, str_dst, str_edge_type, chunks, label, loss_f

# feature fusion
if self.num_feature == 3:
# feat = torch.cat((out_[utt_idx], hidden_rgcn[utt_idx],
# relatt_out[utt_idx], symbolic_repr), dim=-1)

feat_ = torch.stack([out_[utt_idx], hidden_rgcn[utt_idx], symbolic_repr], dim=1).unsqueeze(2)
feat = self.CoAtt(feat_).squeeze(1).squeeze(1)
output = torch.log_softmax(self.linear(self.ac_tanh(self.dropout(self.linear_out(feat)))), dim=1)

else:

# if self.use_layer_norm:
# output = torch.log_softmax(self.linear(self.ac_tanh(self.layer_norm(self.dropout(self.fusion(feat))))), dim=1)
# else:
# output = torch.log_softmax(self.linear(self.ac_tanh(self.dropout(self.fusion(feat)))), dim=1)

elif self.num_feature == 4:
feat_l = torch.cat((out_[utt_idx], hidden_rgcn[utt_idx], symbolic_repr), dim=-1)

feat_ = torch.stack([out_[utt_idx], hidden_rgcn[utt_idx], symbolic_repr], dim=1).unsqueeze(2)
feat = self.CoAtt(feat_).squeeze(1).squeeze(1)

feat_x = torch.cat((feat_l, feat), dim=-1)

output = torch.log_softmax(self.linear(self.ac_tanh(self.dropout(self.linear_out(feat_x)))), dim=1)

# if self.use_layer_norm:
# output = torch.log_softmax(self.linear(self.ac_tanh(self.layer_norm(self.dropout(self.fusion(feat))))), dim=1)
# else:
# output = torch.log_softmax(self.linear(self.ac_tanh(self.dropout(self.fusion(feat)))), dim=1)
elif self.num_feature == 2:
# feat = torch.cat((out_[utt_idx], hidden_rgcn[utt_idx]), dim=-1)

feat_ = torch.stack([out_[utt_idx], hidden_rgcn[utt_idx]], dim=1).unsqueeze(2)
feat = self.CoAtt(feat_).squeeze(1).squeeze(1)

output = torch.log_softmax(self.linear(self.ac(self.dropout(self.linear_out(feat)))), dim=1)

# if self.use_layer_norm:
# output = torch.log_softmax(self.linear(self.ac_tanh(self.layer_norm(self.dropout(self.fusion(feat))))), dim=1)
# else:
# output = torch.log_softmax(self.linear(self.ac_tanh(self.dropout(self.fusion(feat)))), dim=1)
elif self.num_feature == 1:
feat = out_[utt_idx]
if self.use_layer_norm:
output = torch.log_softmax(self.linear(self.ac_tanh(self.layer_norm(self.dropout(self.fusion(feat))))), dim=1)
else:
output = torch.log_softmax(self.linear(self.ac_tanh(self.dropout(self.fusion(feat)))), dim=1)
else:
# feat = out_[utt_idx] + hidden_rgcn[utt_idx] + relatt_out[utt_idx] + symbolic_repr
feat = out_[utt_idx] + hidden_rgcn[utt_idx] + symbolic_repr
if self.use_layer_norm:
output = torch.log_softmax(self.linear_2(self.ac_tanh(self.layer_norm(self.dropout(self.fusion_2(feat))))), dim=1)
Expand Down