259 MB

Ctrl+K

1 contributor

3D full finetuning: NDHWC train_graph.json + torch_params.json, structured weight_map to_onnx, 3D runtime; 2D spike peak_mb corrected

3331842 verified 4 days ago

finetune
3D full finetuning: NDHWC train_graph.json + torch_params.json, structured weight_map to_onnx, 3D runtime; 2D spike peak_mb corrected 4 days ago
manifest.json

7.05 kB
3D full finetuning: NDHWC train_graph.json + torch_params.json, structured weight_map to_onnx, 3D runtime; 2D spike peak_mb corrected 4 days ago
setup_56_nuc_128nm_in116_fp16.onnx

21.5 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in116_fp32.onnx

42.9 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in116_int8w.onnx

10.9 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in116_int8w_fp16.onnx

10.9 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in132_fp16.onnx

21.5 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in132_fp32.onnx

42.9 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in132_int8w.onnx

10.9 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in132_int8w_fp16.onnx

10.9 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in148_fp16.onnx

21.5 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in148_fp32.onnx

42.9 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in148_int8w.onnx

10.9 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago
setup_56_nuc_128nm_in148_int8w_fp16.onnx

10.9 MB
xet

All models: export Conv3d as exact Conv2d over depth-folded batch (~16x faster on onnxruntime-web WebGPU); drop experimental/ 8 days ago