Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add statistics_path parameter to OVWeightQuantizationConfig #1170

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

Add statistics_path parameter to OVWeightQuantizationConfig #1170

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 9 additions & 0 deletions optimum/intel/openvino/configuration.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -427,6 +427,9 @@ class OVWeightQuantizationConfig(OVQuantizationConfigBase):
retained in their original precision without any quantization.
- "int8_sym" stands for 8-bit integer symmetric quantization without zero point.
- "int8_asym" stands for 8-bit integer asymmetric quantization with zero points per each quantization group.
statistics_path (`str`, *optional*):
The directory to save/load NNCF data-aware statistics to/from. Such caching is useful for speeding up
running many data-aware compression experiments on the same model and dataset.
"""

def __init__(
Expand All @@ -449,6 +452,7 @@ def __init__(
processor: Optional[str] = None,
lora_correction: bool = None,
backup_precision: Optional[str] = None,
statistics_path: Optional[str] = None,
**kwargs,
):
super().__init__(
Expand All @@ -470,6 +474,7 @@ def __init__(
self.gptq = gptq
self.lora_correction = lora_correction
self.backup_precision = backup_precision
self.statistics_path = statistics_path
if kwargs.get("weight_format") is not None:
logger.warning(
"The `weight_format` parameter is deprecated and will be removed in optimum-intel v1.24.0. "
Expand Down Expand Up @@ -611,6 +616,9 @@ def to_nncf_dict(self) -> Dict[str, Any]:
awq = True if self.quant_method == OVQuantizationMethod.AWQ else None
sensitivity_metric = nncf.SensitivityMetric(self.sensitivity_metric) if self.sensitivity_metric else None
backup_mode = nncf.BackupMode(self.backup_precision) if self.backup_precision else None
advanced_parameters = None
if self.statistics_path is not None:
advanced_parameters = nncf.AdvancedCompressionParameters(statistics_path=self.statistics_path)
result = {
"mode": mode,
"ratio": self.ratio,
Expand All @@ -624,6 +632,7 @@ def to_nncf_dict(self) -> Dict[str, Any]:
"gptq": self.gptq,
"lora_correction": self.lora_correction,
"backup_mode": backup_mode,
"advanced_parameters": advanced_parameters,
}
return result

Expand Down
Loading