Move float8_opaque_tensor to prototype

[Xia-Weiwen]

As the title

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/3365

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 20c6579 with merge base 257d18a ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[jerryzh168]

looks great, thanks @Xia-Weiwen , can you help move these two int4 tensors as well:

int4_opaque_tensor.py and int4_plain_int32_tensor.py

[jerryzh168]

@Xia-Weiwen if you mark this as bc-breaking, you'll need to write a bc-breaking note

I think it's fine to skip as well since it's probably not widely used yet since it's relatively new, but adding a bc-breaking note sounds fine as well.

[Xia-Weiwen]

@Xia-Weiwen if you mark this as bc-breaking, you'll need to write a bc-breaking note

I think it's fine to skip as well since it's probably not widely used yet since it's relatively new, but adding a bc-breaking note sounds fine as well.

Thanks for the heads up. To make things easier, let me just change the label to "not user facing".

[Xia-Weiwen]

looks great, thanks @Xia-Weiwen , can you help move these two int4 tensors as well:

int4_opaque_tensor.py and int4_plain_int32_tensor.py

Sure.

[Xia-Weiwen]

looks great, thanks @Xia-Weiwen , can you help move these two int4 tensors as well:

int4_opaque_tensor.py and int4_plain_int32_tensor.py

Hi @liangan1 Could you please take int4_plain_int32_tensor.py ? Thanks.

[liangan1]

@jerryzh168 int4 for xpu should be some mature features, I am not sure why we need to move it into the prototype? int4_plain_int32_tensor.py

[jerryzh168]

@liangan1 the context is that we are trying to make sure what is recommended and supported in torchao very clear, currently some of the APIs from our team we can have commitment to maintaining and we expect to see wide adoption as shown in https://github.com/pytorch/ao#stable-workflows, but we are unsure about others, so we want to clean this up and move all the other flows to prototype first. We can discuss how to graduate from prototype as a separate step. I also don't think there is a clearly defined process to graduate from prototype today. cc @vkuzo @danielvegamyhre

I'll discuss with the team offline to see what is expected here if you'd like to keep this as a stable feature