| fix(export_atc): fix export failed
* Caused by a previous bugfix for overriding "is_ascend_om_enabled"
* And is no longer needed after refector of inference_service
Signed-off-by: YidaHao <haoyida@huawei.com>
| 7 天前 |
| feat(model_utils): add ONNX export and loss compare nodes
Add two utility nodes for LeRobot ACT policy evaluation:
1. ExportOnnxNode - Convert PyTorch policies to ONNX format
- Dynamic batch size support
- Dictionary input wrapper for observation format
- ONNX simplification using onnx-simplifier
- Configurable via ROS2 parameters
2. LossCompareNode - Compare model outputs and compute L1 loss
- generate_target mode: Generate target outputs from batch data
- compute_loss mode: Compute L1 loss between predictions and targets
- Batch processing with progress tracking
- Auto policy path detection
Files added:
- export_onnx_node.py: ONNX export node
- loss_compare_node.py: Loss comparison node
- setup.py: Package configuration
- package.xml: ROS2 package manifest
Usage:
ros2 run model_utils export_onnx_node --ros-args \
-p policy_path:=<path> -p device:=cuda
ros2 run model_utils loss_compare_node --ros-args \
-p mode:=compute_loss -p batch_path:=<path>
| 2 个月前 |
| inference_service: generalize compiled wrappers
Split compiled inference into adapter and runtime layers.
Keep backend execution separate from model-family semantics.
Resolve Ascend OM artifacts through config.om.json.
Generate config.om.json from the ATC export path.
Expose backend_type beside policy_type for diagnostics.
Signed-off-by: XiaoqiangWu <wuxiaoqiang.rtos@huawei.com>
| 9 天前 |
| feat(model_utils): add ONNX export and loss compare nodes
Add two utility nodes for LeRobot ACT policy evaluation:
1. ExportOnnxNode - Convert PyTorch policies to ONNX format
- Dynamic batch size support
- Dictionary input wrapper for observation format
- ONNX simplification using onnx-simplifier
- Configurable via ROS2 parameters
2. LossCompareNode - Compare model outputs and compute L1 loss
- generate_target mode: Generate target outputs from batch data
- compute_loss mode: Compute L1 loss between predictions and targets
- Batch processing with progress tracking
- Auto policy path detection
Files added:
- export_onnx_node.py: ONNX export node
- loss_compare_node.py: Loss comparison node
- setup.py: Package configuration
- package.xml: ROS2 package manifest
Usage:
ros2 run model_utils export_onnx_node --ros-args \
-p policy_path:=<path> -p device:=cuda
ros2 run model_utils loss_compare_node --ros-args \
-p mode:=compute_loss -p batch_path:=<path>
| 2 个月前 |
| feat(model_utils): add ONNX export and loss compare nodes
Add two utility nodes for LeRobot ACT policy evaluation:
1. ExportOnnxNode - Convert PyTorch policies to ONNX format
- Dynamic batch size support
- Dictionary input wrapper for observation format
- ONNX simplification using onnx-simplifier
- Configurable via ROS2 parameters
2. LossCompareNode - Compare model outputs and compute L1 loss
- generate_target mode: Generate target outputs from batch data
- compute_loss mode: Compute L1 loss between predictions and targets
- Batch processing with progress tracking
- Auto policy path detection
Files added:
- export_onnx_node.py: ONNX export node
- loss_compare_node.py: Loss comparison node
- setup.py: Package configuration
- package.xml: ROS2 package manifest
Usage:
ros2 run model_utils export_onnx_node --ros-args \
-p policy_path:=<path> -p device:=cuda
ros2 run model_utils loss_compare_node --ros-args \
-p mode:=compute_loss -p batch_path:=<path>
| 2 个月前 |
| fix(model_utils): fix 3 bugs in model utils
* 1. remove exec cmd in ros setup.py
* 2. remove redundant concat in export 3403 onnx path
* 3. make sure --device takes effect when comparing loss
Signed-off-by: YidaHao <haoyida@huawei.com>
| 1 个月前 |
