Vladimir Blagojevic
Vladimir Blagojevic
**Is your feature request related to a problem? Please describe.** It would be great, if possible, to further improve read performance of raw encoded datasets and their subsequent conversion to...
**Describe the bug** Our recent changes for batching pipeline run support have caused a bug in TranslationWrapperPipeline when a wrapped pipeline is QuestionAnswerGenerationPipeline. This particular use case is demoed in...
Starting with the 1.7.0 release Pytorch has added `set_to_none` [keyword](https://pytorch.org/docs/stable/generated/torch.optim.Optimizer.zero_grad.html) to Optimizer's zero_grad method. It seems apex optimizers already implement this functionality but it would be great if they respected...
**Is your feature request related to a problem? Please describe.** We omit the installation of https://github.com/rusty1s/pytorch_scatter in Haystack docker images. PyTorch Scatter is needed for Table QA, and its installation...
**Is your feature request related to a problem? Please describe.** We currently run gunicorn as a root user, exposing not only the contents of the docker image to potential attackers...
**Describe the bug** MostSimilarDocumentsPipeline, unlike all other pipelines, doesn't have the underlying pipeline property/field. As a consequence, any method from the superclass that replies on pipeline (get_type, save_to_yaml) will fail....
### Related Issues - fixes #3210 ### Proposed Changes: During a recent language and tokenization [refactoring](https://github.com/deepset-ai/haystack/issues/2704), we broke FarmReader ONNX conversion support. A simple ONNX conversion of a QA model...
**Describe the bug** During a recent language and tokenization [refactoring](https://github.com/deepset-ai/haystack/issues/2704) we broke FarmReader ONNX conversion support. **Error message** A simple ONNX conversion of a QA model fails; we don't have...
**Describe the bug** HF transformers, as of 4.21.2, support the use of a torch.device objects in their pipeline classes. However, for some reason, they don't support non-indexed `torch.device("cuda")` device, which...
### What? This PR improves the unit tests for OpenAIInvocationLayer focusing on token limitation triggering (and not) ### Why? A user raised an issue in [#5184](https://github.com/deepset-ai/haystack/issues/5184), which appears to be...