riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-31 18:22:34 +06:00
Code Issues Actions 4 Packages Projects Releases Wiki Activity

Also fix loss function issue with the double head models

Browse Source
This commit is contained in:
Catalin Voss 2019-03-24 14:35:55 -07:00
parent 472857c47f
commit 0dd796e359
2 changed files with 8 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
pytorch_pretrained_bert/modeling_gpt2.py
View File

@ -698,8 +698,11 @@ class GPT2DoubleHeadsModel(GPT2PreTrainedModel):
mc_logits = self.multiple_choice_head(hidden_states, mc_token_ids) mc_logits = self.multiple_choice_head(hidden_states, mc_token_ids)
losses = [] losses = []
if lm_labels is not None: if lm_labels is not None:
shift_logits = lm_logits[:, :-1]
shift_labels = lm_labels[:, 1:]
loss_fct = CrossEntropyLoss(ignore_index=-1) loss_fct = CrossEntropyLoss(ignore_index=-1)
losses.append(loss_fct(lm_logits.view(-1, lm_logits.size(-1)), lm_labels.view(-1))) losses.append(loss_fct(shift_logits.view(-1,
shift_logits.size(-1)), shift_labels.view(-1)))
if mc_labels is not None: if mc_labels is not None:
loss_fct = CrossEntropyLoss() loss_fct = CrossEntropyLoss()
losses.append(loss_fct(mc_logits.view(-1, mc_logits.size(-1)), mc_labels.view(-1))) losses.append(loss_fct(mc_logits.view(-1, mc_logits.size(-1)), mc_labels.view(-1)))

5
pytorch_pretrained_bert/modeling_openai.py
View File

@ -811,8 +811,11 @@ class OpenAIGPTDoubleHeadsModel(OpenAIGPTPreTrainedModel):
mc_logits = self.multiple_choice_head(hidden_states, mc_token_ids) mc_logits = self.multiple_choice_head(hidden_states, mc_token_ids)
losses = [] losses = []
if lm_labels is not None: if lm_labels is not None:
shift_logits = lm_logits[:, :-1]
shift_labels = lm_labels[:, 1:]
loss_fct = CrossEntropyLoss(ignore_index=-1) loss_fct = CrossEntropyLoss(ignore_index=-1)
losses.append(loss_fct(lm_logits.view(-1, lm_logits.size(-1)), lm_labels.view(-1))) losses.append(loss_fct(shift_logits.view(-1,
shift_logits.size(-1)), shift_labels.view(-1)))
if mc_labels is not None: if mc_labels is not None:
loss_fct = CrossEntropyLoss() loss_fct = CrossEntropyLoss()
losses.append(loss_fct(mc_logits.view(-1, mc_logits.size(-1)), mc_labels.view(-1))) losses.append(loss_fct(mc_logits.view(-1, mc_logits.size(-1)), mc_labels.view(-1)))