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

[gpt2] fix generation tests (#35822)

Browse Source 
fix gpt2 generation tests
This commit is contained in:
Joao Gante
2025-01-22 09:41:04 +00:00
committed by GitHub
parent f439e28d32
commit 36c9181f5c
1 changed files with 10 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
tests/models/gpt2/test_modeling_gpt2.py
View File

@@ -651,16 +651,18 @@ class GPT2ModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin
outputs = model.generate( outputs = model.generate(
input_ids=input_ids, input_ids=input_ids,
attention_mask=inputs["attention_mask"].to(torch_device), attention_mask=inputs["attention_mask"].to(torch_device),
max_length=20,
) )
outputs_tt = model.generate( outputs_tt = model.generate(
input_ids=input_ids, input_ids=input_ids,
attention_mask=inputs["attention_mask"].to(torch_device), attention_mask=inputs["attention_mask"].to(torch_device),
token_type_ids=token_type_ids, token_type_ids=token_type_ids,
max_length=20,
) )
inputs_non_padded = tokenizer(sentences[0], return_tensors="pt").input_ids.to(torch_device) inputs_non_padded = tokenizer(sentences[0], return_tensors="pt").input_ids.to(torch_device)
output_non_padded = model.generate(input_ids=inputs_non_padded) output_non_padded = model.generate(input_ids=inputs_non_padded, max_length=20)
num_paddings = inputs_non_padded.shape[-1] - inputs["attention_mask"][-1].long().sum().cpu().item() num_paddings = inputs_non_padded.shape[-1] - inputs["attention_mask"][-1].long().sum().cpu().item()
inputs_padded = tokenizer(sentences[1], return_tensors="pt").input_ids.to(torch_device) inputs_padded = tokenizer(sentences[1], return_tensors="pt").input_ids.to(torch_device)
@@ -711,16 +713,18 @@ class GPT2ModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin
outputs = model.generate( outputs = model.generate(
input_ids=input_ids, input_ids=input_ids,
attention_mask=inputs["attention_mask"].to(torch_device), attention_mask=inputs["attention_mask"].to(torch_device),
max_length=20,
) )
outputs_tt = model.generate( outputs_tt = model.generate(
input_ids=input_ids, input_ids=input_ids,
attention_mask=inputs["attention_mask"].to(torch_device), attention_mask=inputs["attention_mask"].to(torch_device),
token_type_ids=token_type_ids, token_type_ids=token_type_ids,
max_length=20,
) )
inputs_non_padded = tokenizer(sentences[0], return_tensors="pt").input_ids.to(torch_device) inputs_non_padded = tokenizer(sentences[0], return_tensors="pt").input_ids.to(torch_device)
output_non_padded = model.generate(input_ids=inputs_non_padded) output_non_padded = model.generate(input_ids=inputs_non_padded, max_length=20)
num_paddings = inputs_non_padded.shape[-1] - inputs["attention_mask"][-1].long().sum().cpu().item() num_paddings = inputs_non_padded.shape[-1] - inputs["attention_mask"][-1].long().sum().cpu().item()
inputs_padded = tokenizer(sentences[1], return_tensors="pt").input_ids.to(torch_device) inputs_padded = tokenizer(sentences[1], return_tensors="pt").input_ids.to(torch_device)
@@ -776,7 +780,7 @@ class GPT2ModelLanguageGenerationTest(unittest.TestCase):
# The dog was found in a field near the intersection of West and West Streets.\n\nThe dog # The dog was found in a field near the intersection of West and West Streets.\n\nThe dog
expected_output_ids = [464, 3290, 373, 1043, 287, 257, 2214, 1474, 262, 16246, 286, 2688, 290, 2688, 27262, 13, 198, 198, 464, 3290,] # fmt: skip expected_output_ids = [464, 3290, 373, 1043, 287, 257, 2214, 1474, 262, 16246, 286, 2688, 290, 2688, 27262, 13, 198, 198, 464, 3290,] # fmt: skip
output_ids = model.generate(input_ids, do_sample=False) output_ids = model.generate(input_ids, do_sample=False, max_length=20)
if verify_outputs: if verify_outputs:
self.assertListEqual(output_ids[0].tolist(), expected_output_ids) self.assertListEqual(output_ids[0].tolist(), expected_output_ids)
@@ -805,13 +809,13 @@ class GPT2ModelLanguageGenerationTest(unittest.TestCase):
torch.manual_seed(0) torch.manual_seed(0)
tokenized = tokenizer("Today is a nice day and", return_tensors="pt", return_token_type_ids=True) tokenized = tokenizer("Today is a nice day and", return_tensors="pt", return_token_type_ids=True)
input_ids = tokenized.input_ids.to(torch_device) input_ids = tokenized.input_ids.to(torch_device)
output_ids = model.generate(input_ids, do_sample=True) output_ids = model.generate(input_ids, do_sample=True, max_length=20)
output_str = tokenizer.decode(output_ids[0], skip_special_tokens=True) output_str = tokenizer.decode(output_ids[0], skip_special_tokens=True)
token_type_ids = tokenized.token_type_ids.to(torch_device) token_type_ids = tokenized.token_type_ids.to(torch_device)
output_seq = model.generate(input_ids=input_ids, do_sample=True, num_return_sequences=5) output_seq = model.generate(input_ids=input_ids, do_sample=True, num_return_sequences=5, max_length=20)
output_seq_tt = model.generate( output_seq_tt = model.generate(
input_ids=input_ids, token_type_ids=token_type_ids, do_sample=True, num_return_sequences=5 input_ids=input_ids, token_type_ids=token_type_ids, do_sample=True, num_return_sequences=5, max_length=20
) )
output_seq_strs = tokenizer.batch_decode(output_seq, skip_special_tokens=True) output_seq_strs = tokenizer.batch_decode(output_seq, skip_special_tokens=True)
output_seq_tt_strs = tokenizer.batch_decode(output_seq_tt, skip_special_tokens=True) output_seq_tt_strs = tokenizer.batch_decode(output_seq_tt, skip_special_tokens=True)