SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Remove my unhelpful comments :)

Browse Source
This commit is contained in:
Catalin Voss
2019-03-27 10:45:11 -07:00
parent fda2f62395
commit 01520d5412
2 changed files with 2 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
pytorch_pretrained_bert/modeling_gpt2.py
View File

@@ -621,9 +621,7 @@ class GPT2LMHeadModel(GPT2PreTrainedModel):
shift_logits = lm_logits[:, :-1].contiguous()
shift_labels = lm_labels[:, 1:].contiguous()
# In tensorflow, it's [batch, d_0, d_1, ..., d_{r-1}, num_classes]
# in pytorch, it's [batch, num_classes, d_0, d_1, ..., d_{r-1}]
# We just flatten the tokens out this way.
# Flatten the tokens
loss_fct = CrossEntropyLoss(ignore_index=-1)
loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)),
shift_labels.view(-1))

4
pytorch_pretrained_bert/modeling_openai.py
View File

@@ -720,9 +720,7 @@ class OpenAIGPTLMHeadModel(OpenAIGPTPreTrainedModel):
shift_logits = lm_logits[:, :-1].contiguous()
shift_labels = lm_labels[:, 1:].contiguous()
# In tensorflow, it's [batch, d_0, d_1, ..., d_{r-1}, num_classes]
# in pytorch, it's [batch, num_classes, d_0, d_1, ..., d_{r-1}]
# We just flatten the tokens out this way.
# Flatten the tokens
loss_fct = CrossEntropyLoss(ignore_index=-1)
loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)),
shift_labels.view(-1))