Hello,

Trying the fine-tuning example and trying to load the data, I get an error:

hasura=# SELECT pgml.load_dataset('kde4', kwargs => '{"lang1": "en", "lang2": "es"}');ERROR:  functionpgml.load_dataset(unknown, kwargs=> unknown) does not existLINE1:SELECTpgml.load_dataset('kde4', kwargs=>'{"lang1": "en", ...               ^HINT:  No function matches the given name and argument types. You might need to add explicit type casts.hasura=#\df pgml.load*                                                   List of functions Schema |     Name     |          Result data type           |               Argument data types                | Type--------+--------------+-------------------------------------+--------------------------------------------------+------ pgml   | load_dataset | TABLE(table_name text, rows bigint) | source text, "limit" bigint DEFAULT NULL::bigint | func(1 row)

Is there something else that needs to be done to load these sample data? This is for pgml14 and postgresql-14 on Ubuntu 22

Fine tuning is the last 1.0 (Python) feature that has not been ported to 2.0 (Rust) yet. If you want to start fine tuning now, you'll need to install1.0. 

Some work has been completed to move the core huggingface functionality, (e.g.tokenizers) to Rust, but there is plenty of "task" specific logic to implement for all the various different fine tuning strategies. That part isn't particularly daunting (we could just copy/paste the 1.0 Python into 2.0), but I'm more interested in creating a generic API (implemented in Rust) so that users can define their own fine tuning/text/image/audio/video processing pipelines (beyond pre-built model specific ones already included) via SQL in 2.0.