riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-08-01 18:51:14 +06:00
Code Issues Actions 3 Packages Projects Releases Wiki Activity

[TFGPT2] - Fix flaky past_key_values test (#9460)

Browse Source 
* fix tf flakey

* remove test files
This commit is contained in:
Patrick von Platen 2021-01-07 16:12:08 +01:00 committed by GitHub
parent 758ed3332b
commit f33a6f3446
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 7 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
tests/test_modeling_tf_gpt2.py
View File

@ -218,6 +218,11 @@ class TFGPT2ModelTester:
): ):
model = TFGPT2Model(config=config) model = TFGPT2Model(config=config)
input_ids = input_ids[:1, :]
input_mask = input_mask[:1, :]
token_type_ids = token_type_ids[:1, :]
self.batch_size = 1
# first forward pass # first forward pass
outputs = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, use_cache=True) outputs = model(input_ids, attention_mask=input_mask, token_type_ids=token_type_ids, use_cache=True)
@ -225,13 +230,13 @@ class TFGPT2ModelTester:
# create hypothetical next token and extent to next_input_ids # create hypothetical next token and extent to next_input_ids
next_tokens = ids_tensor((self.batch_size, 3), config.vocab_size) next_tokens = ids_tensor((self.batch_size, 3), config.vocab_size)
next_token_types = ids_tensor((self.batch_size, 3), self.type_vocab_size)
next_attn_mask = ids_tensor((self.batch_size, 3), 2) next_attn_mask = ids_tensor((self.batch_size, 3), 2)
next_token_types = ids_tensor((self.batch_size, 3), self.type_vocab_size)
# append to next input_ids and token_type_ids # append to next input_ids and token_type_ids
next_input_ids = tf.concat([input_ids, next_tokens], axis=-1) next_input_ids = tf.concat([input_ids, next_tokens], axis=-1)
next_token_type_ids = tf.concat([token_type_ids, next_token_types], axis=-1)
next_attention_mask = tf.concat([input_mask, next_attn_mask], axis=-1) next_attention_mask = tf.concat([input_mask, next_attn_mask], axis=-1)
next_token_type_ids = tf.concat([token_type_ids, next_token_types], axis=-1)
output_from_no_past = model( output_from_no_past = model(
next_input_ids, token_type_ids=next_token_type_ids, attention_mask=next_attention_mask next_input_ids, token_type_ids=next_token_type_ids, attention_mask=next_attention_mask