ayo/ipex-llm
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

Nano: update optimizer's examples (#6917)

Browse source 
* update resnet example

* update

* fix style

* update

* update vision transformer

* fix ut
This commit is contained in:
Ruonan Wang 2022-12-09 09:51:38 +08:00 committed by GitHub
parent 7a8bd4cee5
commit d90ee7e51e
1 changed files with 13 additions and 15 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

28
README.md
View file

@ -178,22 +178,20 @@ The output of `optimizer.summary()` will be something like:
-------------------------------- ---------------------- -------------- ----------------------
| method | status | latency(ms) | accuracy |
-------------------------------- ---------------------- -------------- ----------------------
| original | successful | 43.688 | 0.969 |
| fp32_ipex | successful | 33.383 | not recomputed |
| bf16 | fail to forward | None | None |
| bf16_ipex | early stopped | 203.897 | None |
| int8 | successful | 10.74 | 0.969 |
| jit_fp32 | successful | 38.732 | not recomputed |
| jit_fp32_ipex | successful | 35.205 | not recomputed |
| jit_fp32_ipex_channels_last | successful | 19.327 | not recomputed |
| openvino_fp32 | successful | 10.215 | not recomputed |
| openvino_int8 | successful | 8.192 | 0.969 |
| onnxruntime_fp32 | successful | 20.931 | not recomputed |
| onnxruntime_int8_qlinear | successful | 8.274 | 0.969 |
| onnxruntime_int8_integer | fail to convert | None | None |
| original | successful | 45.145 | 0.975 |
| bf16 | successful | 27.549 | 0.975 |
| static_int8 | successful | 11.339 | 0.975 |
| jit_fp32_ipex | successful | 40.618 | 0.975* |
| jit_fp32_ipex_channels_last | successful | 19.247 | 0.975* |
| jit_bf16_ipex | successful | 10.149 | 0.975 |
| jit_bf16_ipex_channels_last | successful | 9.782 | 0.975 |
| openvino_fp32 | successful | 22.721 | 0.975* |
| openvino_int8 | successful | 5.846 | 0.962 |
| onnxruntime_fp32 | successful | 20.838 | 0.975* |
| onnxruntime_int8_qlinear | successful | 7.123 | 0.981 |
-------------------------------- ---------------------- -------------- ----------------------
Optimization cost 64.3s in total.
* means we assume the precision of the traced model does not change, so we don't recompute accuracy to save time.
Optimization cost 60.8s in total.
```
</details>