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 bnb ground truth (#39117)

Browse Source 
* update bnb resulte

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

* set seed to avoid sampling different results

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

* fix int8 tests

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

* fix typo

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

* add comments

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

---------

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>
This commit is contained in:
jiqing-feng 2025-07-02 02:06:37 +08:00 committed by GitHub
parent 260846efad
commit db2f535443
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 10 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
tests/quantization/bnb/test_4bit.py
View File

@ -27,6 +27,7 @@ from transformers import (
AutoTokenizer,
BitsAndBytesConfig,
pipeline,
set_seed,
)
from transformers.models.opt.modeling_opt import OPTAttention
from transformers.testing_utils import (
@ -111,6 +112,8 @@ class Base4bitTest(unittest.TestCase):
EXPECTED_OUTPUTS.add("Hello my name is John Doe, I am a student at the University")
EXPECTED_OUTPUTS.add("Hello my name is John and I am 25 years old.")
EXPECTED_OUTPUTS.add("Hello my name is John and I am a student at the University of")
# Expected values on Intel XPU and NV A100
EXPECTED_OUTPUTS.add("Hello my name is Alina. I have been working as a professional")
MAX_NEW_TOKENS = 10
def setUp(self):
@ -513,6 +516,8 @@ class Pipeline4BitTest(Base4bitTest):
max_new_tokens=self.MAX_NEW_TOKENS,
)
# Avoid sampling different outputs
set_seed(42)
# Real second forward pass
pipeline_output = self.pipe(self.input_text)
self.assertIn(pipeline_output[0]["generated_text"], self.EXPECTED_OUTPUTS)

5
tests/quantization/bnb/test_mixed_int8.py
View File

@ -27,6 +27,7 @@ from transformers import (
AutoTokenizer,
BitsAndBytesConfig,
pipeline,
set_seed,
)
from transformers.models.opt.modeling_opt import OPTAttention
from transformers.testing_utils import (
@ -113,6 +114,8 @@ class BaseMixedInt8Test(unittest.TestCase):
MAX_NEW_TOKENS = 10
# Expected values with offload
EXPECTED_OUTPUTS.add("Hello my name is John and I am a professional photographer based in")
# Expected values on Intel XPU and NV A100
EXPECTED_OUTPUTS.add("Hello my name is Alina. I have been working as a professional")
def setUp(self):
# Models and tokenizer
@ -649,6 +652,8 @@ class MixedInt8TestPipeline(BaseMixedInt8Test):
max_new_tokens=self.MAX_NEW_TOKENS,
)
# Avoid sampling different outputs
set_seed(42)
# Real second forward pass
pipeline_output = self.pipe(self.input_text)
self.assertIn(pipeline_output[0]["generated_text"], self.EXPECTED_OUTPUTS)