T4rec use case (#853)

* start working on next item prediction usecase

* clean up notebook

* update

* update

* update

* update

* switch to using BERT

* update

* add synthetic data & unit test

* fix unit test

* update

* fix unit test; clean up

* add schema

* update

* add tracking logo

* use absolute path for data

* fix unit test

* fix unit test

* fix paths in test

* Add test for retrieval model with transformer block (#833)

* Add test for transformer with RetrievalModelV2

* Update test for transformer retrieval model

* Remove test_retrieval from test_block

* Allow index param to be optional to `Encoder.encode`

* Correct target extraction in `SequencePredictNext`

* Replace ragged coercion with axis aware tf.squeeze

* Revert change to predict next

* Remove unused ReplaceMaskedEmbeddings (only required for MLM model)

* Support tuple return type from model.fit `pre` argument

* Use predict last and use as pre instead of transform

* Revert changes to contrastive output

* Set process_lists default value to False

* Add d_model and MLPBlock

* Revert change to `Encoder.encode`

* Revert change to default value of `process_lists` in `sample_batch`

* Remove commented query_embeddings line

* update comment about prediction tuple

Co-authored-by: Marc Romeyn <[email protected]>

* Enable concat of sequence features with `InputBlockV2` (#883)

* Add test for seuqence features concat

* Update EmbeddingTable squeeze logic to handle ragged tensors

* Update docs building actions (#884)

- Update to avoid deprecation messages.

- Delete the .doctrees and .buildinfo files
  from the sphinx-multiversion builds.

* update

* Do not copy source to docs output (#888)

Set `html_copy_source` to `False` to
avoid copying the documentation source
files to the output. This should help
reduce the size of the `gh-pages` branch.

Also, run the common workflow that
removes stale review directories. This should
also reduce the size of `gh-pages` and the repo.

* update

* update

* update

Co-authored-by: rnyak <[email protected]>
Co-authored-by: Oliver Holworthy <[email protected]>
Co-authored-by: Marc Romeyn <[email protected]>
Co-authored-by: mikemckiernan <[email protected]>
Co-authored-by: bschifferer <[email protected]>

Author: 6 people