You are viewing a plain text version of this content. The canonical link for it is here.
Posted to discuss-archive@tvm.apache.org by isong via Apache TVM Discuss <no...@discuss.tvm.ai> on 2020/10/06 09:27:24 UTC
[Apache TVM Discuss] [Questions] [VTA] Deploy Pretrained Vision Model
from Pytorch on VTA
Hi everyone working on VTA and @thierry,
I am trying to load Pytorch model and see how the VTA instruction is generated.
Here is the notebook based on the [tutorial](https://tvm.apache.org/docs/vta/tutorials/frontend/deploy_classification.html#sphx-glr-download-vta-tutorials-frontend-deploy-classification-py) and [this discussion](https://discuss.tvm.apache.org/t/pytorch-quantization-error-during-quantization/7167).
- https://gist.github.com/insop/18aaab92d48061c6c0e440be0bd83733
However, at the end I don't see any VTA instruction is running when I tried `sim_stats = simulator.stats()`.
```
Execution statistics:
inp_load_nbytes : 0
wgt_load_nbytes : 0
acc_load_nbytes : 0
uop_load_nbytes : 0
out_store_nbytes: 0
gemm_counter : 0
alu_counter : 0
```
Based on MxNet tutorial mentioned above, I am don't see why there is no VTA instruction is issued ( also enabled `debug_flag_` and don't see debug instruction is printed either).
Can anyone help?
Thank you,
ISS
---
[Visit Topic](https://discuss.tvm.apache.org/t/vta-deploy-pretrained-vision-model-from-pytorch-on-vta/8092/1) to respond.
You are receiving this because you enabled mailing list mode.
To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/4009c35ebe5e714e2e468725c80c60c99ac9fd1ae9e033df223b6f91c07adf1d).
[Apache TVM Discuss] [Questions] [VTA] Deploy Pretrained Vision Model from Pytorch on VTA
Posted by shlrao via Apache TVM Discuss <no...@discuss.tvm.ai>.
It is this question has any process? I meet this situation
---
[Visit Topic](https://discuss.tvm.apache.org/t/vta-deploy-pretrained-vision-model-from-pytorch-on-vta/8092/2) to respond.
You are receiving this because you enabled mailing list mode.
To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/eac5841c13babcf767ea4ad545853d87cfcf5e579379b5855be2b33ffc1a67ae).