Skip to content

Add recipes for using NPU Estimate Latency pass in Olive #196

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

Draft
alinah-amd wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

Add recipes for using NPU Estimate Latency pass in Olive #196

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
39978cd
Add recipes for using NPU Estimate Latency pass in Olive
alinah-amd Dec 8, 2025
13072c9
Merge branch 'main' into alinah/npu_estimator_recipes
alinah-amd Dec 8, 2025
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
138 changes: 138 additions & 0 deletions microsoft-resnet-50/aitk/resnet_qdq_amd_plus_estimate_npu_latency.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,138 @@
{
"input_model": {
"type": "HfModel",
"model_path": "microsoft/resnet-50",
"task": "image-classification",
"io_config": {
"input_names": [
"pixel_values"
],
"input_shapes": [
[
1,
3,
224,
224
]
],
"output_names": [
"logits"
]
}
},
"systems": {
"target_system": {
"type": "LocalSystem"
}
},
"data_configs": [
{
"name": "quantize_data_config",
"type": "HuggingfaceContainer",
"user_script": "imagenet.py",
"load_dataset_config": {
"data_name": "timm/mini-imagenet",
"split": "train",
"streaming": true
},
"pre_process_data_config": {
"type": "dataset_pre_process",
"size": 64,
"cache_key": "imagedata_quantization",
"transpose": true
},
"post_process_data_config": {
"type": "dataset_post_process"
}
},
{
"name": "evaluation_data_config",
"type": "HuggingfaceContainer",
"user_script": "imagenet.py",
"load_dataset_config": {
"data_name": "timm/mini-imagenet",
"split": "validation",
"streaming": true
},
"pre_process_data_config": {
"type": "dataset_pre_process",
"size": 1000,
"cache_key": "imagedata_evaluation",
"transpose": true
},
"post_process_data_config": {
"type": "dataset_post_process"
}
}
],
"evaluators": {
"common_evaluator": {
"metrics": [
{
"name": "accuracy",
"type": "accuracy",
"data_config": "quantize_data_config",
"sub_types": [
{
"name": "accuracy_score",
"priority": 1,
"metric_config": {
"task": "multiclass",
"num_classes": 1001
}
}
]
},
{
"name": "latency",
"type": "latency",
"data_config": "quantize_data_config",
"sub_types": [
{
"name": "avg",
"priority": 2
}
]
}
]
}
},
"passes": {
"conversion": {
"device": "cpu",
"type": "OnnxConversion",
"target_opset": 17,
"save_as_external_data": true,
"all_tensors_to_one_file": true,
"dynamic": false,
"use_dynamo_exporter": false
},
"transpose_input": {
"type": "InputNCHWtoNHWC"
},
"OnnxQuantization": {
"type": "OnnxQuantization",
"data_config": "quantize_data_config",
"activation_type": "uint8",
"precision": "uint8",
"calibrate_method": "MinMax",
"save_as_external_data": true
},
"addmetadata": {
"type": "VitisAIAddMetaData",
"config_meta_data_keys": [
"architectures",
"model_type"
],
"activation_type": "uint8",
"weight_type": "uint8",
"quant_type": "OnnxStaticQuantization"
},
"estimate_npu_latency": { "type": "EstimateNPULatency", "target_device": "stx" }
},
"target": "target_system",
"evaluator": "common_evaluator",
"cache_dir": "cache",
"output_dir": "model/resnet_ptq_qnn",
"evaluate_input_model": false
}
61 changes: 61 additions & 0 deletions timm-mobilenetv3_small_100.lamb_in1k/olive/config_plus_estimate_npu_latency.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,61 @@
{
"input_model": {
"type": "PytorchModel",
"model_path": "timm/mobilenetv3_small_100.lamb_in1k",
"model_loader": "load_timm",
"model_script": "user_script.py",
"io_config": {
"input_names": [ "x" ],
"input_shapes": [ [ 1, 3, 224, 224 ] ],
"output_names": [ "output" ]
}
},
"systems": {
"local_system": {
"type": "LocalSystem"
}
},
"data_configs": [
{
"name": "imagenet1k",
"type": "HuggingfaceContainer",
"user_script": "imagenet.py",
"load_dataset_config": { "data_name": "imagenet-1k", "split": "validation", "streaming": true },
"pre_process_data_config": { "type": "dataset_pre_process", "size": 256, "cache_key": "imagenet256" },
"post_process_data_config": { "type": "imagenet_post_fun" }
}
],
"evaluators": {
"common_evaluator": {
"metrics": [
{
"name": "accuracy",
"type": "accuracy",
"backend": "huggingface_metrics",
"data_config": "imagenet1k",
"sub_types": [
{ "name": "accuracy", "priority": 1, "goal": { "type": "max-degradation", "value": 0.05 } }
]
},
{
"name": "latency",
"type": "latency",
"data_config": "imagenet1k",
"sub_types": [
{ "name": "avg", "priority": 2, "goal": { "type": "percent-min-improvement", "value": 0.1 } },
{ "name": "max" },
{ "name": "min" }
]
}
]
}
},
"passes": {
"conversion": { "type": "OnnxConversion" },
"matmul4": { "type": "OnnxBlockWiseRtnQuantization", "accuracy_level": 4 },
"mnb_to_qdq": { "type": "MatMulNBitsToQDQ" },
"estimate_npu_latency": { "type": "EstimateNPULatency", "target_device": "stx" }
},
"evaluator": "common_evaluator",
"output_dir": "models/timm"
}