mirror of
https://github.com/huggingface/transformers.git
synced 2025-07-04 05:10:06 +06:00
parent
4acf692ace
commit
bb2a44ad4b
@ -1821,7 +1821,10 @@ class QuarkConfig(QuantizationConfigMixin):
|
|||||||
from quark.torch.export.config.config import JsonExporterConfig
|
from quark.torch.export.config.config import JsonExporterConfig
|
||||||
from quark.torch.export.main_export.quant_config_parser import QuantConfigParser
|
from quark.torch.export.main_export.quant_config_parser import QuantConfigParser
|
||||||
from quark.torch.quantization.config.config import Config
|
from quark.torch.quantization.config.config import Config
|
||||||
|
else:
|
||||||
|
raise ImportError(
|
||||||
|
"Quark is not installed. Please refer to https://quark.docs.amd.com/latest/install.html."
|
||||||
|
)
|
||||||
# This might be e.g. `"fp8"` or `"awq"`.
|
# This might be e.g. `"fp8"` or `"awq"`.
|
||||||
self.custom_mode = kwargs["quant_method"]
|
self.custom_mode = kwargs["quant_method"]
|
||||||
self.legacy = "export" not in kwargs
|
self.legacy = "export" not in kwargs
|
||||||
|
@ -33,6 +33,7 @@ if is_quark_available():
|
|||||||
from quark.torch.export.nn.modules.qparamslinear import QParamsLinear
|
from quark.torch.export.nn.modules.qparamslinear import QParamsLinear
|
||||||
|
|
||||||
|
|
||||||
|
@require_quark
|
||||||
class QuarkConfigTest(unittest.TestCase):
|
class QuarkConfigTest(unittest.TestCase):
|
||||||
def test_commmon_args(self):
|
def test_commmon_args(self):
|
||||||
config = AutoConfig.from_pretrained("amd/Llama-3.1-8B-Instruct-w-int8-a-int8-sym-test")
|
config = AutoConfig.from_pretrained("amd/Llama-3.1-8B-Instruct-w-int8-a-int8-sym-test")
|
||||||
|
Loading…
Reference in New Issue
Block a user