riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-03 12:50:06 +06:00
Code Issues Actions 28 Packages Projects Releases Wiki Activity

update gemma tests (#38384)

Browse Source 
* update

* update

* update

* update

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
This commit is contained in:
Yih-Dar 2025-05-26 19:54:04 +02:00 committed by GitHub
parent cd0f3ce73b
commit 07848a8405
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 4 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

31
tests/models/gemma/test_modeling_gemma.py
View File

@ -30,7 +30,6 @@ from transformers.testing_utils import (
require_torch,
require_torch_accelerator,
require_torch_gpu,
require_torch_sdpa,
slow,
torch_device,
)
@ -147,7 +146,7 @@ class GemmaIntegrationTest(unittest.TestCase):
EXPECTED_TEXTS = [
"Hello I am doing a project on the 1990s and I need to know what the most popular music",
"Hi today I am going to share with you a very easy and simple recipe of <strong><em>Khichdi",
"Hi today I am going to share with you a very easy and simple recipe of <strong><em>Kaju Kat",
]
model = AutoModelForCausalLM.from_pretrained(model_id, low_cpu_mem_usage=True, torch_dtype=torch.bfloat16).to(
@ -168,34 +167,12 @@ class GemmaIntegrationTest(unittest.TestCase):
EXPECTED_TEXTS = [
"Hello I am doing a project on the 1990s and I need to know what the most popular music",
"Hi today I am going to share with you a very easy and simple recipe of <strong><em>Khichdi",
"Hi today I am going to share with you a very easy and simple recipe of <strong><em>Kaju Kat",
]
# bfloat16 gives strange values, likely due to it has lower precision + very short prompts
model = AutoModelForCausalLM.from_pretrained(
model_id, low_cpu_mem_usage=True, torch_dtype=torch.bfloat16, attn_implementation="eager"
)
model.to(torch_device)
tokenizer = AutoTokenizer.from_pretrained(model_id)
inputs = tokenizer(self.input_text, return_tensors="pt", padding=True).to(torch_device)
output = model.generate(**inputs, max_new_tokens=20, do_sample=False)
output_text = tokenizer.batch_decode(output, skip_special_tokens=True)
self.assertEqual(output_text, EXPECTED_TEXTS)
@require_torch_sdpa
@require_read_token
def test_model_2b_sdpa(self):
model_id = "google/gemma-2b"
EXPECTED_TEXTS = [
"Hello I am doing a project on the 1990s and I need to know what the most popular music",
"Hi today I am going to share with you a very easy and simple recipe of <strong><em>Khichdi",
]
model = AutoModelForCausalLM.from_pretrained(
model_id, low_cpu_mem_usage=True, torch_dtype=torch.bfloat16, attn_implementation="sdpa"
model_id, low_cpu_mem_usage=True, torch_dtype=torch.float16, attn_implementation="eager"
)
model.to(torch_device)