Get Embeddings

Get all embedding data for 500k (or otherwise requested) data points.

This will take a random sample of the dataset. It exports the result as a csv in order to reduce request size and response latency (as results can be large) This will return:

• id
• gold (index)
• prediction (index)
• data quality
• text_sample
• x coordinate
• y coordinate ** any metadata columns explicitly requested

Set data_view=true to return data embeddings instead of model embeddings. If data embeddings are not available, an exception will be thrown

Set use_seed=false for random batches of embeddings on every response. This is useful for batching, but should remain true if you are requesting a single set of embeddings. If you are returning all embeddings, this parameter is useless