I would like use Relay's built in post-training quantization along with BYOC.
Inside the quantize pass params are bound to the main function which causes an
issue downstream since I need tensors to be VarNode instead of ConstantNode in
BYOC. I think this can be resolved by unbinding params af
Thanks @matt-arm! Will be trying out some patterns today.
---
[Visit
Topic](https://discuss.tvm.ai/t/external-codegen-status-of-annotating-composite-functions/6150/8)
to respond.
You are receiving this because you enabled mailing list mode.
To unsubscribe from these emails, [click
here]
Hi @masahi thanks for bringing this to my attention. Looks like this PR could
work for us too. As a first pass we hope to target the most common fusion
patterns as in your PR.
---
[Visit
Topic](https://discuss.tvm.ai/t/external-codegen-status-of-annotating-composite-functions/6150/7)
to
While post-training quantization from float32 to int8 hidden/cell states
remains an open research topic, one work around we've found is to compute
hidden states at higher precision on CPU rather than on the low-precision
accelerator in order to reach our accuracy requirements.
>From a fronte
Thank you @comaniac and @matt-arm. I will see if MergeCompilerRegions is
something that will work for us, otherwise will keep an eye out for your PR.
---
[Visit
Topic](https://discuss.tvm.ai/t/external-codegen-status-of-annotating-composite-functions/6150/4)
to respond.
You are receiving
I was playing around with the new external codegen & composite functions and
noticed this flow fails inside of AnnotateTarget. Wondering if it is because of
this
[TODO](https://github.com/apache/incubator-tvm/blob/021213832cb98703dda54f631215ac17fbabff7b/src/relay/transforms/annotate_target.cc
@xwrock For now I think the proper way to work around this is to go the BYOC
route until this part of TVM becomes more flexible for custom accelerators.
---
[Visit
Topic](https://discuss.tvm.ai/t/tensorize-how-to-use-tensorize-for-composition-op-eg-conv-relu/2336/8)
to respond.
You are r
