Aug 12 15:03 megaloc: update README — CoreML multifunction batch_1/2/4/8 usage, measured benchmarks 15:02 megaloc: CoreML INT8 multifunction package (batch_1/2/4/8, float32 I/O, output 'descriptor') 15:02 megaloc: CoreML FP32 multifunction package (batch_1/2/4/8, float32 I/O, output 'descriptor') Aug 9 14:04 fix(mixvpr): re-export ALL models with the trained checkpoint loaded Previous onnx/coreml exports never loaded resnet50_MixVPR_4096_channels(1024)_rows(4).ckpt — they shipped an ImageNet backbone with a randomly initialized aggregator, and the CosSim checks passed because the PyTorch reference was the same unloaded model. Symptom: same-image 90%-crop cosine ~0.48 (should be ~0.77); indoor relocalization recall@2m 59.5% -> 94.8% after the fix.
Export scripts now load the ckpt inside the model constructor with a strict key check and refuse to run without it. Jul 15 01:39 Update benchmark with full 6-model comparison 01:37 Update CoreML INT8 (fixed batch, fast) 01:37 Update CoreML FP16 (fixed batch) 01:37 Update CoreML FP32 (fixed batch, fast) 01:37 Update ONNX INT8 (multi-batch) 01:37 Add ONNX FP32 weights (multi-batch) 01:36 Update ONNX FP32 (multi-batch) Jul 14 19:38 Update README with onnx/coreml structure 19:37 Delete megaloc/megaloc_int8.onnx with huggingface_hub 19:37 Delete megaloc/megaloc_fp16.onnx with huggingface_hub 19:37 Delete megaloc/megaloc_coreml_int8.mlpackage/Manifest.json with huggingface_hub 19:37 Delete megaloc/megaloc_coreml_int8.mlpackage/Data/com.apple.CoreML/weights/weight.bin with huggingface_hub 19:37 Delete megaloc/megaloc_coreml_int8.mlpackage/Data/com.apple.CoreML/model.mlmodel with huggingface_hub 19:37 Delete megaloc/megaloc_coreml_fp32.mlpackage/executorch_debug_handle_mapping.json with huggingface_hub 19:37 Delete megaloc/megaloc_coreml_fp32.mlpackage/Manifest.json with huggingface_hub 19:37 Delete megaloc/megaloc_coreml_fp32.mlpackage/Data/com.apple.CoreML/weights/weight.bin with huggingface_hub 19:37 Delete megaloc/megaloc_coreml_fp32.mlpackage/Data/com.apple.CoreML/model.mlmodel with huggingface_hub 19:37 Delete megaloc/megaloc.onnx with huggingface_hub 19:37 Delete megaloc/README.md with huggingface_hub 19:36 Add CoreML INT8 (multi-batch) 19:36 Add CoreML FP16 (multi-batch) 19:35 Add CoreML FP32 (multi-batch) 19:34 Add ONNX INT8 19:34 Add ONNX FP16 (multi-batch) 19:34 Move ONNX FP32 to onnx/ Jul 7 17:13 Add megaloc README with benchmarks and usage 17:12 Add CoreML FP32 model 17:11 Add CoreML INT8 model 17:11 Upload mixvpr/README.md with huggingface_hub 17:10 Add ONNX INT8 model 17:09 Add ONNX FP16 model 17:09 Upload mixvpr/coreml/mixvpr_int8.mlpackage/Data/com.apple.CoreML/weights/weight.bin with huggingface_hub 17:08 Upload mixvpr/coreml/mixvpr_int8.mlpackage/Data/com.apple.CoreML/model.mlmodel with huggingface_hub 17:08 Upload mixvpr/coreml/mixvpr_int8.mlpackage/Manifest.json with huggingface_hub 17:08 Upload mixvpr/coreml/mixvpr_fp16.mlpackage/Data/com.apple.CoreML/weights/weight.bin with huggingface_hub 17:08 Upload mixvpr/coreml/mixvpr_fp16.mlpackage/Data/com.apple.CoreML/model.mlmodel with huggingface_hub 17:08 Upload mixvpr/coreml/mixvpr_fp16.mlpackage/Manifest.json with huggingface_hub 17:08 Upload mixvpr/onnx/mixvpr_fp32.onnx with huggingface_hub 17:07 Upload mixvpr/onnx/mixvpr_fp16.onnx with huggingface_hub 17:07 Add ONNX FP32 model 17:02 initial commit 
