diff options
| author | John Bauer <horatio@gmail.com> | 2022-07-30 09:36:04 +0300 |
|---|---|---|
| committer | John Bauer <horatio@gmail.com> | 2022-08-14 11:21:19 +0300 |
| commit | 194e446c200e7c57d01901748c60324be877d6e8 (patch) | |
| tree | 125b2b83c8b20b0e12c0f66c12dc26a9a88112c0 | |
| parent | 91e9e9c2cd27a8278a6dab4bfec8d0e46b5681be (diff) | |
Just to compare masked or not-masked for xlnetmasks2
| -rw-r--r-- | stanza/models/common/bert_embedding.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/stanza/models/common/bert_embedding.py b/stanza/models/common/bert_embedding.py index 49cd72f3..af7f685f 100644 --- a/stanza/models/common/bert_embedding.py +++ b/stanza/models/common/bert_embedding.py @@ -216,7 +216,7 @@ def extract_bert_embeddings(model_name, tokenizer, model, data, device, keep_end features = [] for i in range(int(math.ceil(len(data)/128))): with torch.no_grad(): - attention_mask = torch.tensor(tokenized['attention_mask'][128*i:128*i+128], device=device) + attention_mask = None id_tensor = torch.tensor(tokenized['input_ids'][128*i:128*i+128], device=device) feature = model(id_tensor, attention_mask=attention_mask, output_hidden_states=True) # feature[2] is the same for bert, but it didn't work for |