mathislucka

Extend support for token classification

1

**Is your feature request related to a problem? Please describe.** It would be great if the support for token classification could be extended beyond what the `Extractor` currently offers. Specifically,...

Training of FARMReader uses too many and potentially wrong no answer labels due to bug in SquadProcessor

3

**Describe the bug** When training a QA model with the `FARMReader.train()` method, the `SquadProcessor` will be used to convert the SQuAD-style json file to training samples. The context for a...

Support M1 GPU in FARMReader

6

**Is your feature request related to a problem? Please describe.** Since haystack v1.6 we have support for pytorch 1.12 which also means support for the M1 GPU. However, we currently...

**Describe the bug** When using the `PreProcessor` with the `remove_substrings` parameter with a configuration where the PreProcessor should remove substrings with a space in the string, pipeline validation fails if...

FARMReader should convert local hf-transformers model to FARM if necessary

1

**Proposed changes**: Before this change, only remote models in huggingface transformers format could be used by the FARMReader. Sometimes, people might have stored these locally and still want to use...

add sql-like query strings for more convenience

2

This implements a method allowing filters to be constructed from query strings. It should mainly be more convenient and quicker to type for less complex filters. Query strings can be...

Add OptimumEmbedder

2

**Is your feature request related to a problem? Please describe.** Huggin Face's Optimum library provides faster inference through ONNX and TensorRT. This can be used to create blazing fast embedding...

We'll soon integrate prompts from prompt-hub into dC. One of the features we want to offer is sorting the prompts by creation date. For that, it would be very nice...