Add a python example on how to upload and use OPT

[JoaoLages]

First of all, thank you for open-sourcing OPT! 💪🚀

I'd like to ask for a simple example of how to use the new OPT model.
In here there could be a python code description on how to upload the model or, because probably the code is too big, point to a script with that example/tutorial.

[spencer-hong]

Hi @JoaoLages, I'm not associated with Facebook, but based on looking through the repository, Facebook provides a script to consolidate your shards here. https://github.com/facebookresearch/metaseq/blob/main/metaseq/distributed/stitch_fsdp_ckpt.py

[nickums]

Eager to use this amazing product, after following the setup instructions, I am looking for a way to start using it.
First, having selected a pre-trained model to use (reshard-model_part-0.pt etc),
what directory do i store them in, or, what load procedure should I use?
Second, what script to use for a simple 'prompt and response' interaction?
presumably using megatron/text-generation-server.py

[suchenzang]

Consolidating all asks for model loading into #88, which will be released soon from HuggingFace.

We also have #78 and #77 open to fix some of these usability issues on our end.

[nickums]

Dear Susan, thankyou for your response I am now working through issues 26, 19, 57, 77, 78, 88 Nick

…

On Thu, 12 May 2022 at 06:43, Susan Zhang ***@***.***> wrote: Consolidating all asks for model loading into #88 <#88>, which will be released soon from HuggingFace. We also have #78 <#78> and #77 <#77> open to fix some of these usability issues on our end. — Reply to this email directly, view it on GitHub <#56 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADYVMFLCPCEOGRHAYASPROTVJSLADANCNFSM5VIY5NXA> . You are receiving this because you commented.Message ID: ***@***.***>

-- Nick "Mac" McElwaine