haystack-core-integrations

haystack-core-integrations copied to clipboard

MistralChatAdapter's ALLOWED_PARAMS has `max_tokens` but generation_kwargs is using `max_gen_len`. Because of this discrepancy, `max_tokens` limit is set to the default `512` whenever running the model. - Haystack version: 2.1.2 -...

vish9812

Allow vanilla qdrant filters

3

Allow passing a qdrant filter object instead of a haystack filter object - this is useful when the haystack filter object does not support the use case, e.g. when to...

jdb78

align tests and docs on NVIDIA_API_KEY (instead of NVIDIA_CATALOG_API_KEY)

4

README.md says `Note: integration tests will be skipped unless the env var NVIDIA_API_KEY is set.` with this MR, the statement is now true for all cases.

mattf

adding support of "amazon.titan-embed-text-v2:0"

1

Adding support of newly launched amazon.titan-embed-text-v2:0. Implemented and tested, working fine.

jdixosnd

make hosted nim default

1

all inference calls should be directed to integrate.api.nvidia.com. the nvcf inference is deprecated and will be removed. this allows for users to get on integrate.api.nvidia.com now and avoid any disruptions.

mattf

**Is your feature request related to a problem? Please describe.** Llama CPP generators don't support streaming. **Describe the solution you'd like** Implement support for streaming, as done for other generators...

anakin87

All examples showing integration of Qdrant with haystack show how to create a new Qdrant index/collection and use that as a DocumentStore. However, there does not seem to be a...

miary

Fixes: https://github.com/deepset-ai/haystack-core-integrations/issues/655 Added support for custom mapping when creating a new index using ElasticsearchDocumentStore. Tested the fix using the below code: ``` from haystack_integrations.document_stores.elasticsearch import ElasticsearchDocumentStore from haystack import Pipeline...

AnushreeBannadabhavi

llama.cpp: LlamaCppGenerator.run() raises 'TypeError' when passing {"stream": True}

2

**Describe the bug** When calling `LlamaCppGenerator.run()` with `generation_kwargs={"stream": True}`, a TypeError "'generator' object is not subscriptable" is raised in line 97: `replies = [output["choices"][0]["text"]]`, because the `create_completion` function of the...

paulgekeler

**Is your feature request related to a problem? Please describe.** In `InMemoryEmbeddingRetriever` there is the parameter `return_embedding` that returns also embedding in `Document`, with OpenSearchEmbeddingRetriever I must recalculates embeddings after...

nickprock