* enable Opensearch unit tests under Win
* move unit tests into a dedicated job
* skip audio tests on missing dependencies
* avoid failing test collection when soundfile is not available
* Update .github/workflows/tests.yml
Co-authored-by: Sara Zan <sara.zanzottera@deepset.ai>
Co-authored-by: Sara Zan <sara.zanzottera@deepset.ai>
* fixed tokens in question generation
* simplified assignment
* same behavior also for pad and eos
* use skip_special_tokens in batch_decode
* fixed black error and update docs
* fixed schemas ci error
* JSON schemas
* Add git diff to debug schema issues
* opensearch schema was missing
* Add missing instruction in the workflow error message
* typo
* Ability to run Ray Serve detached
Fixes#2944
Ability to run Ray Serve detached - to allow running multiple instances of the app (HA).
See https://docs.ray.io/en/latest/serve/package-ref.html#core-apis
* Generating the docs
* Re-trigger the CI pipeline
* Retrigger the CI Pipeline
* Typo in docstrings
* Fixing docstring and typing issues
* Regenerating docs
* [EMPTY] Re-trigger CI
* [EMPTY] Re-trigger CI
* Refactoring to allow any number of args for the `serve.start()` method
There seems to be additional arguments of the `serve.start()` method, so we should probably cover all of them at once, instead of only the `detached` option.
* [EMPTY] Re-trigger CI
* Test whether the ServeControllerClient in fact has the supplied `detached` parameter
* Tutorial 06: Replace DPR with EmbeddingRetriever
Closes#2887
* Add updated tutorials/6.md file
Replace `DensePassageRetriever` with `EmbeddingRetriever`
* Update Tutorial 06 based on PR feedback
* Further updates to Tutorial-06 according to review feedback
* [Tutorial 06] Put in review feedback for the py file
* Extending the Ray Serve integration to allow attributes for Serve deployments
This closes#2917
We should be able to set Ray Serve attributes for the nodes of pipelines, like amount of GPU to use, max_concurrent_queries, etc.
Now this is possible from the pipeline yaml file for each node of the pipeline.
* Ran black and regenerated the json schemas
* Fixing the JSON Schema generation
* Trying to fix the schema CI test issue
* Fixing the test and the schemas
Python 3.8 was generating a different schema than Python 3.7 is creating in the CI. You MUST use Python 3.7 to generate the schemas, otherwise the CIs will fail.
* Merge the two Ray pipeline test cases
* Generate the JSON schemas again after `$ pip install .[all]`
* Removing `haystack/json-schemas/haystack-pipeline-1.16.schema.json`
This was generated by the JSON generator, but based on @ZanSara's instructions, I am removing it.
* Making changes based on @ZanSara's request - the newly requested test is failing
* Fixing the JSON schema generation again
* Renaming `replicas` and moving it under `serve_deployment_kwargs`
* add extras validation, untested
* Dcoumentation update
* Black
* [EMPTY] Re-trigger CI
Co-authored-by: Sara Zan <sarazanzo94@gmail.com>
* Update minimum requirement for selenium for using the crawler
* Updating pin of grpcio to match default in google colab
* Adding requests requirement
* Bug fix Weaviate document deletion
If no filters param is passed in, then the original code retrieves *all* documents before then deleting by their IDs. There's no need for that, since we can delete by their IDs directly.
* Edit comment to clarify deletion and recreation
* Write unit tests for bug fix
* Bugfix: batch_size was not passed to self.generate_batch
* Testing pre-push hooks
* Formatting code using black
* Adding black changes
* Adding black changes
* Adding black changes
Co-authored-by: ZanSara <sarazanzo94@gmail.com>
* add Opensearch extras
* let OpenSearchDocumentStore use opensearch-py
* Update Documentation & Code Style
* fix a bug found after adding tests
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Sara Zan <sara.zanzottera@deepset.ai>
* Upgrading Weaviate used for testing to 1.14.1 from 1.11.0
This has also brought up an issue with one of the test filtering for value "a". This test has started to fail, as "a" is a default stopword in Weaviate, so I have changed this test to look for value "c" instead of value "a" to get around the stopword issue.
* Weaviate client upgrade
From v3.3.3 to v3.6.0
* Adding BM25 Retrieval to Weaviate
Weaviate now supports BM25 retrieval in experiment mode and with some limitations (like it cannot be combined with filters).
This commit adds support for inverted index (BM25) querying against Weaviate.
* Running Black on the recent code changes
* Update Documentation & Code Style
* Fixing linting issues after code changes by black
* The BM25 query needs to be in all lowercase for now
The BM25 query needs to be provided all lowercase while the functionality is in experimental mode in Weaviate.
See https://app.slack.com/client/T0181DYT9KN/C017EG2SL3H/thread/C017EG2SL3H-1658790227.208119
* Fixing method parameter docstring to highlight that they are not supported in Weaviate
* Update Documentation & Code Style
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Add pre-commit config
* update contributing guidelines
* try failing the workflow
* add pre-commit to the deps
* updating uninstall instructions
* separate jobs in CI
* make tutorials check fail
* make black check fail
* make openapi check fail
* make yaml schema and api docs checks fail
* highlight the instructions
* Update .pre-commit-config.yaml
Co-authored-by: Tobias Wochinger <mail@tobias-wochinger.de>
* Update CONTRIBUTING.md
Co-authored-by: Tobias Wochinger <mail@tobias-wochinger.de>
* Update CONTRIBUTING.md
Co-authored-by: Tobias Wochinger <mail@tobias-wochinger.de>
* Use black --check
* Add images of the CI
* title level
* feedback
Co-authored-by: Tobias Wochinger <mail@tobias-wochinger.de>
* restart tutorials in the loop
* remove container steps in tutorials.yml
* forgotten quotes
* unmatched bracket
* give names to containers
* try to limit the log size
* make the containers restart on the scripts as well
* feedback
* Raise integration tests timeout
* raising limit again
* Add support for model folder into BasePreProcessor
* First draft of custom model on PreProcessor
* Update Documentation & Code Style
* Update tests to support custom models
* Update Documentation & Code Style
* Test for wrong models in custom folder
* Default to ISO names on custom model folder
Use long names only when needed
* Update Documentation & Code Style
* Refactoring language names usage
* Update fallback logic
* Check unpickling error
* Updated tests using parametrize
Co-authored-by: Sara Zan <sara.zanzottera@deepset.ai>
* Refactored common logic
* Add format control to NLTK load
* Tests improvements
Add a sample for specialized model
* Update Documentation & Code Style
* Minor log text update
* Log model format exception details
* Change pickle protocol version to 4 for 3.7 compat
* Removed unnecessary model folder parameter
Changed logic comparisons
Co-authored-by: Sara Zan <sara.zanzottera@deepset.ai>
* Update Documentation & Code Style
* Removed unused import
* Change errors with warnings
* Change to absolute path
* Rename sentence tokenizer method
Co-authored-by: tstadel
* Check document content is a string before process
* Change to log errors and not warnings
* Update Documentation & Code Style
* Improve split sentences method
Co-authored-by: Sara Zan <sara.zanzottera@deepset.ai>
* Update Documentation & Code Style
* Empty commit - trigger workflow
* Remove superfluous parameters
Co-authored-by: tstadel
* Explicit None checking
Co-authored-by: tstadel
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Sara Zan <sara.zanzottera@deepset.ai>
* clean up tests and run earlier
* use change detection
* better naming, skip ES
* more cleanup
* fix job name
* dummy commit to trigger the CI
* mock away the PDF converter
* make the test compatible with 3.7
* removed leftover
* always run the api tests, use a matrix for the OS
* refactor all the tests
* remove outdated dependency
* pylint
* new abstract method
* adjust for older python versions
* rename pipeline file
* address PR comments
