Skip to content

Quantization with ONNX #20

@hbandukw

Description

@hbandukw

location of model after dynamic quantization with ONNX

https://www.kaggle.com/models/hbandukw/distilbertforclimatedisinfo_quantized

script location

Hina_dev/Quantization/DynamicQuantization_onnx.ipynb

Log:

Missing keys: []
Unexpected keys: []
✅ Model exported to distilbert_model.onnx
⚠️ Warning: Static quantization failed. Falling back to dynamic quantization.
✅ Dynamic quantized model saved at distilbert_quantized_dynamic.onnx
⚡ Starting ONNX evaluation...
📝 Processing batch 1...
✔️ Batch 1: 30/32 correct
📝 Processing batch 2...
✔️ Batch 2: 31/32 correct
📝 Processing batch 3...
✔️ Batch 3: 32/32 correct
📝 Processing batch 4...
✔️ Batch 4: 31/32 correct
📝 Processing batch 5...
✔️ Batch 5: 31/32 correct
📝 Processing batch 6...
✔️ Batch 6: 32/32 correct
📝 Processing batch 7...
✔️ Batch 7: 32/32 correct
📝 Processing batch 8...
✔️ Batch 8: 32/32 correct
📝 Processing batch 9...
✔️ Batch 9: 31/32 correct
📝 Processing batch 10...
✔️ Batch 10: 30/32 correct
📝 Processing batch 11...
✔️ Batch 11: 30/32 correct
📝 Processing batch 12...
✔️ Batch 12: 30/32 correct
📝 Processing batch 13...
✔️ Batch 13: 31/32 correct
📝 Processing batch 14...
✔️ Batch 14: 30/32 correct
📝 Processing batch 15...
✔️ Batch 15: 31/32 correct
📝 Processing batch 16...
✔️ Batch 16: 28/32 correct
📝 Processing batch 17...
✔️ Batch 17: 30/32 correct
📝 Processing batch 18...
✔️ Batch 18: 31/32 correct
📝 Processing batch 19...
✔️ Batch 19: 30/32 correct
📝 Processing batch 20...
✔️ Batch 20: 31/32 correct
📝 Processing batch 21...
✔️ Batch 21: 31/32 correct
📝 Processing batch 22...
✔️ Batch 22: 32/32 correct
📝 Processing batch 23...
✔️ Batch 23: 32/32 correct
📝 Processing batch 24...
✔️ Batch 24: 32/32 correct
📝 Processing batch 25...
✔️ Batch 25: 30/32 correct
📝 Processing batch 26...
✔️ Batch 26: 31/32 correct
📝 Processing batch 27...
✔️ Batch 27: 32/32 correct
📝 Processing batch 28...
✔️ Batch 28: 31/32 correct
📝 Processing batch 29...
✔️ Batch 29: 32/32 correct
📝 Processing batch 30...
✔️ Batch 30: 29/32 correct
📝 Processing batch 31...
✔️ Batch 31: 28/32 correct
📝 Processing batch 32...
✔️ Batch 32: 29/32 correct
📝 Processing batch 33...
✔️ Batch 33: 31/32 correct
📝 Processing batch 34...
✔️ Batch 34: 30/32 correct
📝 Processing batch 35...
✔️ Batch 35: 31/32 correct
📝 Processing batch 36...
✔️ Batch 36: 32/32 correct
📝 Processing batch 37...
✔️ Batch 37: 30/32 correct
📝 Processing batch 38...
✔️ Batch 38: 30/32 correct
📝 Processing batch 39...
✔️ Batch 39: 2/3 correct
ONNX Quantized Model Accuracy: 0.9590 (1169/1219)

Metadata

Metadata

Assignees

Labels

No labels
No labels

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions