Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

Can you check the result for .predict and .score of gemini-2.0-flash-lite-001? If no quota, they shouldn't have meaningful output.

Please actively followup with BQML, when the quota issue is resolved, then to add fine tuning and other missing stuff.

Can you check the result for .predict and .score of gemini-2.0-flash-lite-001? If no quota, they shouldn't have meaningful output.

Here is the output when running .score() for gemini-2.0-flash-lite.
message=Internal error in Dremel: INTERNAL: An internal error has occurred.; Unable to find default quota for model name: gemini-2.0-flash-lite. debug=; code: \t INTERNAL_ERROR\ndescription: \"An internal error has occurred.; Unable to find default quota for model name: gemini-2.0-flash-lite-001
Note: this test is blocked by b/413131111

Output when running .predict() for gemini-2.0-flash-lite.
E google.api_core.exceptions.BadRequest: 400 GET https://bigquery.googleapis.com/bigquery/v2/projects/bigframes-load-testing/queries/bdfcd82f-4e59-412d-81e0-2f8d3e411aa8?maxResults=0&location=US&prettyPrint=false: Option(s) DATA_SPLIT_METHOD, MAX_ITERATIONS, INPUT_LABEL_COLS, PROMPT_COL are not supported for GEMINI_1P5_FLASH model training.
Note: this bug is an internal bug, BQML is working on rolling out this feature.