[microTVM] AOT Demo #8075
[microTVM] AOT Demo #8075
Conversation
mehrdadh
force-pushed
the
aot_zephyr
branch
2 times, most recently
from
15b61d2
to
a397a6b
Compare
apps/microtvm/zephyr/demo_runtime/boards/nrf5340dk_nrf5340_cpuapp.conf
Outdated
Show resolved
Hide resolved
|
@areusch tested this on NRF board, however, the model doesn't fit on Nucleo board. So I removed the boards that has not been tested so far in the aot_demo. |
| lowered = relay.build(relay_mod, target, params=params) | ||
|
|
||
| # Load sample and generate input/output header files | ||
| sample = np.load(os.path.join(this_dir, "testdata", "ic_sample_fp32_8.npy")) |
There was a problem hiding this comment.
311KB is quite a bit for this repo...i'm not sure whether we should load from a different place?
There was a problem hiding this comment.
replaced it with this link to its source:
image classification model
There was a problem hiding this comment.
thanks, also possible to do so for this one?
There was a problem hiding this comment.
For the sample data I need to use a sample that has predictable result to test in the CI. I got this sample from a large dataset and need to keep it somewhere. The other option is to create a personal repository and add it there. We could reuse that repo later for other other samples as well. Let me know what you think.
Also I can save this numpy array as int8 and reduce the size 4 times if that works.
* initial * remove compare * temp fix * debugging * hack * hack for testing * both test pass * cleanup * fix tests and tutorials * restructure * cleanup * cleanup * fix check files * fixed for physical devices * address comments * reduce nrf stack size * update sample url * format
* initial * remove compare * temp fix * debugging * hack * hack for testing * both test pass * cleanup * fix tests and tutorials * restructure * cleanup * cleanup * fix check files * fixed for physical devices * address comments * reduce nrf stack size * update sample url * format
This PR makes these changes: