[Hardware][Intel] Support compressed-tensor W8A8 for CPU backend (#7257)

This commit is contained in:
Li, Jiang
2024-09-11 09:46:46 -07:00
committed by GitHub
parent 3b7fea770f
commit 0b952af458
18 changed files with 686 additions and 43 deletions
+3 -2
View File
@@ -95,8 +95,9 @@ def _get_quantization_config(
"""Get the quantization config."""
if model_config.quantization is not None:
quant_config = get_quant_config(model_config, load_config)
if not current_platform.is_tpu():
capability = current_platform.get_device_capability()
capability = current_platform.get_device_capability() # type: ignore
if capability is not None:
capability = capability[0] * 10 + capability[1]
if capability < quant_config.get_min_capability():
raise ValueError(