haystack-core-integrations

haystack-core-integrations copied to clipboard

## Summary and motivation [Chonkie](https://github.com/chonkie-inc/chonkie) is a chunking library. With [release 1.2](https://github.com/chonkie-inc/chonkie/releases/tag/v1.2.0), Chonkie comes with [8 chunkers](https://github.com/chonkie-inc/chonkie?tab=readme-ov-file#chunkers) that could all be made available in a Haystack integration. In particular, SemanticChunker...

julian-risch

Chroma - The meta in write_documents only supports str, int, float, bool, but not list. However, its search filter does support list.

1

#DUG Describe When using ChromaDocumentStore from Haystack integrations, the write_documents method does not support storing list values in metadata fields, only supporting str, int, float, and bool types. However, during...

King-pand

We have expanded `StreamingChunk` to include new fields such as ```python content: str meta: Dict[str, Any] = field(default_factory=dict, hash=False) component_info: Optional[ComponentInfo] = field(default=None) # All fields below here are new...

sjrl

Refer to parent issue for more details

sjrl

No score and good results on PgVector when using cosine_similarity because of the DESC and the 1 - cosine_distance. Wrong usage of order by

3

**Describe the bug** When using the PgvectorDocumentStore as stated in [the documentation of pgvector](https://github.com/pgvector/pgvector?tab=readme-ov-file#troubleshooting) : > The query needs to have an ORDER BY and LIMIT, and the ORDER BY...

plopezamaya

Original issue in Haystack: https://github.com/deepset-ai/haystack/issues/9505 Example implementation in Haystack: https://github.com/deepset-ai/haystack/pull/9693 (using `dataclass.replace`) Potentially affected integrations: - [ ] Bedrock - [ ] Cohere - [ ] FastEmbed - [ ]...

anakin87

Refer to the parent issue for details. Relevant docs: https://openrouter.ai/docs/use-cases/reasoning-tokens

anakin87

Refer to the parent issue for details. Relevant docs: https://docs.mistral.ai/capabilities/reasoning/ To be investigated. Might require major adaptations.

anakin87

Haystack's `finish_reason` is used both in `StreamingChunk` and `ChatMessage.meta["finish_reason"]` to indicate why the text generation stopped. Google GenAI uses `STOP` as a general-purpose finish reason, for both normal text completions...

anakin87

## Summary and motivation S3 Vectors is the latest KB solution from AWS. Although in preview, this would be a great addition to the existing DocStore integrations. ## Detailed design...

jlonge4