* Handle no answers and no documents scenarios in '_process_request'
* Fix tests
* Change return type in '_process_request'
* Return to use dicts
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Add type check for meta & add tests
* Improve tests
* Handle properly the ValueError ad an HTTPException
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Apply filter in eval only if no gold documents are given as input
* change type annotation of input documents in eval
* Update Documentation & Code Style
* fix mypy
* Update Documentation & Code Style
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Add DELETE /feedback for testing and make the ID generate server-side
* Make sure to delete only user generated labels
* Reduce fixture scope, was too broad
* Make test a bit more generic
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* join node should allow reciprocal rank fusion
* Update Documentation & Code Style
* add missing merging mode
* tuples are immutable
* take correct results from pipeline
* Update Documentation & Code Style
* Simple docstrings, use ValueError
* Use K=60
* Minor refactoring
* precalculate expected result in test
* Update Documentation & Code Style
* refactor to make more clear
* rm unused imports
* tests should test only one thing
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: dmigo <d.f.goryunov@gmail.com>
* Fix dependency issue in Tutorial 6
* Remove faiss from first install block
* move faiss group back to main installation step
* Comment out Milvus cell
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* move commandline args to global conftest
* correct test exclude paths
* Update Documentation & Code Style
* exclude test_generator_pipeline_with_translator from windows ci
* exclude further oom tests
* enable log_cli
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Make mypy check also ui and rest_api, fix ui
* Remove explicit type packages from extras, mypy now downloads them
* Make pylint and mypy run on every file except tests
* Rename tasks
* Change cache key
* Fix mypy errors in rest_api
* Normalize python versions to avoid cache misses
* Add all exclusions to make pylint pass
* Run mypy on rest_api and ui as well
* test if installing the package really changes outcome
* Comment out installation of packages
* Experiment: randomize tests
* Add fallback installation steps on cache misses
* Remove randomization
* Add comment on cache
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Fix a path issue in Dockerfile-GPU
* Fix paths in Dockerfile-GPU
* Add workflow_dispatch to docker build task
* Remove reference to optional component from ui/, not needed anymore
* Move pytorch installation last to avoid replacing it later
* Remove optional import from rest_api too, no more needed
* Change path in ui/Dockerfile
* ui container works again
* Complete review of import paths
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* pass documents as extra param to eval
* pass documents via labels to eval
* rename param in docs
* Update Documentation & Code Style
* Revert "rename param in docs"
This reverts commit 2f4c2ec79575e9dd33a8300785f789a327df36f4.
* Revert "pass documents via labels to eval"
This reverts commit dcc51e41f2637d093d81c7d193b873c17c36b174.
* simplify iterating through labels and docs
* Update Documentation & Code Style
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Speed up query_by_embedding in InMemoryDocumentStore.
* Make sure query and document embeddings are of the same dtype since they can vary.
* Handle cases where there are 0 and 1 documents.
* Don't put entire embedding matrix on GPU at once. Use separate get_score
functions for the CPU and GPU.
* Norm the vectors in get_scores_numpy in a safer way.
* Apply Black
* Incorporate missing factor of 4 in memory use calculation.
* Apply Black
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Upgrade pydoc-markdown and fix the YAMLs to work with it
* Pin pydoc-markdown to major version
* Generalize pydoc-markdown workflow
* Make a single Action to perform all tasks that require committing into the local branch
* Merge the code updates and the docs in the Linux CI to prevent the bot from always show the pipeline as green
* Installing Jupyter deps for Black
* Build cache before running generation tasks
* Add check not to run the code generation on master
* Simplify push action
* Add more test deps in setup.cfg and remove from GH Action workflow
* Remove forced upgrades on pip install
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Conversion to df does not need initialization
* Apply Black
* fix test case
* Apply Black
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* add filters attribute to labels and use in eval
* Add latest docstring and tutorial changes
* overwrite params if None
* populate filters from Label to MultiLabel
* add query_id in eval df and deepcopy params for each label
* fix mypy
* add test for aggregating filters in multilabel
* use query ids also in answers df
* loop through unique query_ids
* hash filters and query text as id
* Add latest docstring and tutorial changes
* fix top_k reader eval
* Apply Black
* rename query_id to id/multilabel_id
* Apply Black
* json dump filters in dataframe
* add filters and id to wrong_examples()
* Apply Black
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Sara Zan <sara.zanzottera@deepset.ai>
* Testing black on ui/
* Applying black on docstores
* Add latest docstring and tutorial changes
* Create a single GH action for Black and docs to reduce commit noise to the minimum, slightly refactor the OpenAPI action too
* Remove comments
* Relax constraints on pydoc-markdown
* Split temporary black from the docs. Pydoc-markdown was obsolete and needs a separate PR to upgrade
* Fix a couple of bugs
* Add a type: ignore that was missing somehow
* Give path to black
* Apply Black
* Apply Black
* Relocate a couple of type: ignore
* Update documentation
* Make Linux CI run after applying Black
* Triggering Black
* Apply Black
* Remove dependency, does not work well
* Remove manually double trailing commas
* Update documentation
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* Make FileTypeClassifier more flexible
* Make supported_types a init parameter
* Add tests and fix a couple of bugs
* Formatting
* Fix mypy
* Implement feedback
* Adding simple setup.py to ui/ and rest_api and remove respective extras from main setup.cfg
* Make 'pip install rest_api/' fetch the local Haystack instead of downloading from pypi
* Add some comments to the new setup.py files and fix the Dockerfiles
* Add version info to 'farm-haystack-ui'
* Fix the OpenAPI Specs workflow
* Install rest_api and ui properly on the CI too
* Make the workflow see changes on every setup file
* Fix workflow cache keys
* Add license to rest_api and ui
* Revert "Make the docstring bot work only on master (#2078)"
This reverts commit 649d07405770cd59696d0120107a3b2f0aafe7c2.
* Add latest docstring and tutorial changes
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* 🎨 Update type annotations to allow their extraction for JSON Schema
* ✨ Add main script doing all the work to generate the JSON Schema
* ➕ Add GitHub Action dependency to generate JSON Schema
* ✨ Update JSON Schema generation script to allow easily generating the schema without making a PR
* 👷 Add GitHub Action to generate JSON Schema
* 💚 Fix CI GitHub Action
* 💚 Update GitHub Action environment variables
* ✨ Add initial JSON Schema
* Add latest docstring and tutorial changes
* 🐛 Do not allow extra params not defined in each model
* ♻️ Make any additional properties invalid
* ✨ Make other additional properties invalid in all the levels in pipelines
* ♻️ Do not include Base classes as possible nodes
* 🍱 Update JSON Schema
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* provide option to recreate es doc store on initialization
* Add latest docstring and tutorial changes
* Label expects more arguments
* Label expects also an answer
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Sara Zan <sara.zanzottera@deepset.ai>
* distribute tinybert loss calculation
* improve doc string
* undo unnecessary change
* fix for only one gpu
* adding type hints
* making sure model distillation still works without gpu
* fix bug
* fixing type hints
* Review changes
* Added the synonym analyser for search fields
* Added the review requests.
* Added the synonyms the OpenSearchDocumentStore and review requests.
* Disable cache on the CI
* Reintroduce paths
* Add most files to the cache key
* remove date and path from cache key
* Try double install with cache
* Try to cache more stuff, on a per-commit basis
* Fix windows CI too
* Add comment on how to speed up the CI with better caching
