riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-08-02 19:21:31 +06:00
Code Issues Actions 3 Packages Projects Releases Wiki Activity

Generate: move misplaced test (#29902)

Browse Source
This commit is contained in:
Joao Gante 2024-04-01 12:45:25 +01:00 committed by GitHub
parent e4f5b57a3b
commit c9f6e5e351
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 13 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
tests/generation/test_utils.py
View File

@ -1467,17 +1467,6 @@ class GenerationTesterMixin:
past_kv[i][1].shape, (batch_size, num_attention_heads, seq_length, per_head_embed_dim)
)
def test_generate_from_inputs_embeds_with_bos_token_id_is_none(self):
article = "Today a dragon flew over Paris."
model = AutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-gpt2").to(torch_device)
tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-gpt2")
input_ids = tokenizer(article, return_tensors="pt").input_ids.to(torch_device)
inputs_embeds = model.get_input_embeddings()(input_ids)
model.generate(inputs_embeds=inputs_embeds, max_length=20, bos_token_id=None)
with self.assertRaises(ValueError):
model.generate(max_length=20, bos_token_id=None)
def test_generate_from_inputs_embeds_decoder_only(self):
# When supported, tests that the decoder model can generate from `inputs_embeds` instead of `input_ids`
# if fails, you should probably update the `prepare_inputs_for_generation` function
@ -2817,3 +2806,16 @@ class GenerationIntegrationTests(unittest.TestCase, GenerationIntegrationTestsMi
self.assertTrue(y_prob > 0.001 and n_prob > 0.001)
self.assertTrue(y_prob <= 1.0 and n_prob <= 1.0)
def test_generate_from_inputs_embeds_with_bos_token_id_is_none(self):
article = "Today a dragon flew over Paris."
model = AutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-gpt2").to(torch_device)
tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-gpt2")
input_ids = tokenizer(article, return_tensors="pt").input_ids.to(torch_device)
inputs_embeds = model.get_input_embeddings()(input_ids)
model.generate(inputs_embeds=inputs_embeds, max_length=20, bos_token_id=None)
# bos_token_id is required when no input ids nor inputs_embeds is passed
with self.assertRaises(ValueError):
model.generate(max_length=20, bos_token_id=None)