ViT-B/16 with Prompt Tuning for image classification. Prompt tokens are prepended to patch embeddings at each transformer block.
iic/cv_vitb16_classification_vision-efficient-tuning-prompt| Item | Value |
|---|---|
| NPU | Ascend910B4 |
| CANN | 8.5.1 |
| PyTorch | 2.9.0+cpu |
| torch_npu | 2.9.0.post1 |
| Python | 3.11.14 |
CPU vs NPU cosine similarity:
| Metric | Value |
|---|---|
| Cosine Similarity | 0.999992 |
Output shape: [1, 100]
Logits sample: [-1.027, -3.788, 0.819, -1.239, -1.530]
Predicted class: 61
SUCCESSpython3 inference.py
python3 evaluate.pyinference.py - NPU inference scriptevaluate.py - CPU vs NPU cosine similarity evaluation