diff --git a/tests/post_training/data/wc_reference_data.yaml b/tests/post_training/data/wc_reference_data.yaml index 91e7e4be220..3d05bc0cc56 100644 --- a/tests/post_training/data/wc_reference_data.yaml +++ b/tests/post_training/data/wc_reference_data.yaml @@ -22,8 +22,8 @@ tinyllama_int8_data_free_backend_TORCH: metric_value: 0.95624 num_int4: 0 num_int8: 312 -tinyllama_data_aware_gptq_backend_OV: - metric_value: 0.86521 +tinyllama_data_aware_gptq_scale_estimation_stateful_backend_OV: + metric_value: 0.86697 num_int4: 94 num_int8: 124 metrics_xfail_reason: "Issue-148819" @@ -43,4 +43,4 @@ tinyllama_NF4_scale_estimation_stateful_per_channel_backend_OV: tinyllama_awq_backup_mode_none_backend_OV: metric_value: 0.84793 num_int4: 208 - num_int8: 0 \ No newline at end of file + num_int8: 0 diff --git a/tests/post_training/model_scope.py b/tests/post_training/model_scope.py index 7ee651723af..ed4b912a5dd 100644 --- a/tests/post_training/model_scope.py +++ b/tests/post_training/model_scope.py @@ -452,7 +452,7 @@ "backends": [BackendType.TORCH], }, { - "reported_name": "tinyllama_data_aware_gptq", + "reported_name": "tinyllama_data_aware_gptq_scale_estimation_stateful", "model_id": "tinyllama/tinyllama-1.1b-step-50k-105b", "pipeline_cls": LMWeightCompression, "compression_params": { @@ -460,7 +460,12 @@ "ratio": 0.8, "mode": CompressWeightsMode.INT4_SYM, "gptq": True, + "scale_estimation": True, + "advanced_parameters": AdvancedCompressionParameters( + scale_estimation_params=AdvancedScaleEstimationParameters(32, 5, 10, 1.0) + ), }, + "params": {"is_stateful": True}, "backends": [BackendType.OV], }, {