Fabian Reinold

JSONReader: Encoding fixes UnicodeDecodeError

1

# Description Provides additional encoding parameter in JSONReader, default is UTF-8. Fixes "UnicodeDecodeError: 'charmap' codec can't decode byte 0x9d in position 3445: character maps to " on Windows 10 x64...

CPU Image: High memory usage on startup

2

### System Info Image: v1.2 CPU Model used: jinaai/jina-embeddings-v2-base-de Deployment: Docker / RH OpenShift ### Information - [X] Docker - [ ] The CLI directly ### Tasks - [X] An...

We currently use Qdrant in conjunction with LangChain to build RAG applications. For a quick sanity check and demo purposes it would be great to get our Markdown documents, which...

Distributed deployment: List snapshots returns no items although they exist on another node

2

I'm trying to build an automated snapshot job running once a day. When bulding the job I only ran one node but since upgraded to 2 nodes with 6 shards...

[Bug]: Langfuse Integration doesn't use custom SSL certficates

4

### What happened? Although specifying a custom SSL certificate, the Langfuse Integration still throws SSLVerifyExceptions. This probably stems from the Langfuse Integration not using the Custom httpx Handlers in which...

### The Feature The proxy configuration provides additional keys for each model: - `input_cost_per_1m_token` - `output_cost_per_1m_token` These are then divided by `1e6` internally to compute the values of the existing...

Token ID computation in BM25/BM42: Possible id overlap due to absolute value casting of hash

1

Found thus only cause I was curious how fastembed calculates token ids for bm25/bm42, so I took a deep dive: ## Problem The [function compute_token_id](https://github.com/qdrant/fastembed/blob/0638c011dc3bc36318b55648e17a5cbb342f1f7d/fastembed/sparse/bm25.py#L282) computes the absolute value of...

Incompability to Python3.13

3

I'm working with the GLiNER library and noticed it lists sentencepiece as a dependency. However, after searching through the codebase, I can't find any place where sentencepiece is actually imported...

Publish official Helm chart to Artifact Hub for discoverability and automatic documentation

1

Planka already provides a Helm chart for Kubernetes deployments, but it is not yet published on [Artifact Hub](https://artifacthub.io/), the central index for Helm charts and Kubernetes packages. Publishing it there...