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

[tests] make quanto tests device-agnostic (#36328)

Browse Source 
* make device-agnostic

* name change
This commit is contained in:
Fanli Lin
2025-02-21 21:20:40 +08:00
committed by GitHub
parent 678885bbbd
commit 7c5bd24ffa
1 changed files with 6 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
tests/quantization/quanto_integration/test_quanto.py
View File

@@ -22,7 +22,6 @@ from transformers.testing_utils import (
require_optimum_quanto, require_optimum_quanto,
require_read_token, require_read_token,
require_torch_accelerator, require_torch_accelerator,
require_torch_gpu,
slow, slow,
torch_device, torch_device,
) )
@@ -181,11 +180,11 @@ class QuantoQuantizationTest(unittest.TestCase):
""" """
self.check_inference_correctness(self.quantized_model, "cpu") self.check_inference_correctness(self.quantized_model, "cpu")
def test_generate_quality_cuda(self): def test_generate_quality_accelerator(self):
""" """
Simple test to check the quality of the model on cuda by comparing the generated tokens with the expected tokens Simple test to check the quality of the model on accelerators by comparing the generated tokens with the expected tokens
""" """
self.check_inference_correctness(self.quantized_model, "cuda") self.check_inference_correctness(self.quantized_model, torch_device)
def test_quantized_model_layers(self): def test_quantized_model_layers(self):
from optimum.quanto import QBitsTensor, QModuleMixin, QTensor from optimum.quanto import QBitsTensor, QModuleMixin, QTensor
@@ -215,7 +214,7 @@ class QuantoQuantizationTest(unittest.TestCase):
) )
self.quantized_model.to(0) self.quantized_model.to(0)
self.assertEqual( self.assertEqual(
self.quantized_model.transformer.h[0].self_attention.query_key_value.weight._data.device.type, "cuda" self.quantized_model.transformer.h[0].self_attention.query_key_value.weight._data.device.type, torch_device
) )
def test_serialization_bin(self): def test_serialization_bin(self):
@@ -430,7 +429,7 @@ class QuantoQuantizationQBitsTensorSerializationTest(QuantoQuantizationSerializa
weights = "int4" weights = "int4"
@require_torch_gpu @require_torch_accelerator
class QuantoQuantizationActivationTest(unittest.TestCase): class QuantoQuantizationActivationTest(unittest.TestCase):
def test_quantize_activation(self): def test_quantize_activation(self):
quantization_config = QuantoConfig( quantization_config = QuantoConfig(
@@ -443,7 +442,7 @@ class QuantoQuantizationActivationTest(unittest.TestCase):
@require_optimum_quanto @require_optimum_quanto
@require_torch_gpu @require_torch_accelerator
class QuantoKVCacheQuantizationTest(unittest.TestCase): class QuantoKVCacheQuantizationTest(unittest.TestCase):
@slow @slow
@require_read_token @require_read_token