riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-08-03 03:31:05 +06:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Fix Seq2seqTrainer decoder attention mask (#26841)

Browse Source 
Don't drop decoder_input_ids without also dropping decoder_attention_mask
This commit is contained in:
Matt 2023-10-18 13:28:15 +01:00 committed by GitHub
parent 280c757f6c
commit 34678db4a1
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/transformers/trainer_seq2seq.py
View File

@ -288,7 +288,9 @@ class Seq2SeqTrainer(Trainer):
and "decoder_input_ids" in generation_inputs
and generation_inputs["labels"].shape == generation_inputs["decoder_input_ids"].shape
):
generation_inputs = {k: v for k, v in inputs.items() if k != "decoder_input_ids"}
generation_inputs = {
k: v for k, v in inputs.items() if k not in ("decoder_input_ids", "decoder_attention_mask")
}
generated_tokens = self.model.generate(**generation_inputs, **gen_kwargs)
# Temporary hack to ensure the generation config is not initialized for each iteration of the evaluation loop