1 Release published by 1 person

• v1.6.1

  published Mar 28, 2025

30 Pull requests merged by 10 people

• Fixing the static-linking.

  #547 merged Mar 29, 2025

• Upgrade candle3

  #545 merged Mar 28, 2025

• Upgrade candle2

  #543 merged Mar 28, 2025

• Moving cublaslt into TEI extension for easier upgrade of candle globally

  #542 merged Mar 28, 2025

• Prepare for release.

  #540 merged Mar 27, 2025

• Fix FromAsCasing warning in Dockerfile-intel

  #541 merged Mar 27, 2025

• Fixing the impure flake devShell to be able to run python code.

  #539 merged Mar 26, 2025

• Add missing match on onnx/model.onnx download

  #472 merged Mar 26, 2025

• make a WA in case Bert model do not have safetensor file

  #515 merged Mar 26, 2025

• Fix VarBuilder handling in GTE e.g. gte-multilingual-reranker-base

  #538 merged Mar 26, 2025

• Small fixup.

  #537 merged Mar 26, 2025

• feat: support HF_ENDPOINT environment when downloading model

  #505 merged Mar 26, 2025

• add CLI flag disable-spans to toggle span trace logging

  #481 merged Mar 26, 2025

• Support classification head for DistilBERT

  #487 merged Mar 26, 2025

• Fixing the tests.

  #531 merged Mar 26, 2025

• Use --hf-token instead of --hf-api-token

  #535 merged Mar 26, 2025

• Add HF_HUB_USER_AGENT_ORIGIN

  #534 merged Mar 26, 2025

• Fusing both Gte Configs.

  #530 merged Mar 26, 2025

• Update README.md to include ONNX

  #507 merged Mar 25, 2025

• feat: add support for "model_type": "gte"

  #519 merged Mar 25, 2025

• chore: Upgrade to tokenizers 0.21.0

  #512 merged Mar 25, 2025

• Fix typo on intel docker image

  #529 merged Mar 25, 2025

• Add intel based images to the CI

  #518 merged Mar 25, 2025

• Fix double incrementing te_request_count metric

  #486 merged Mar 24, 2025

• fix bug for MaskedLanguageModel class`

  #513 merged Mar 17, 2025

• upgrade ipex to 2.6 version for cpu/xpu

  #510 merged Mar 13, 2025

• Optimize flash bert path for hpu device

  #509 merged Mar 11, 2025

• Hpu bucketing

  #489 merged Mar 10, 2025

• Enable splade embeddings for Python backend

  #493 merged Mar 7, 2025

• add TRUST_REMOTE_CODE param to python backend.

  #485 merged Mar 6, 2025

2 Pull requests opened by 2 people

• Refine model file download for python backend

  #526 opened Mar 24, 2025

• Make `sliding_window` for `Qwen2` optional

  #546 opened Mar 28, 2025

7 Issues closed by 4 people

• Support env `HF_ENDPOINT`?

  #416 closed Mar 26, 2025

• Support for DistilBERT Classifier

  #484 closed Mar 26, 2025

• Addition of Snowflake Arctic Embed 2.0

  #444 closed Mar 25, 2025

• Models with `model_type: gte` fail to load, despite the architecture being supported

  #497 closed Mar 25, 2025

• [bug report] Double incrementing te_request_count metric in successful single openai_embed requests

  #480 closed Mar 24, 2025

• Prometheus metrics are empty

  #520 closed Mar 20, 2025

• dify,ragflow添加rerank模型失败

  #516 closed Mar 14, 2025

13 Issues opened by 13 people

• Error occurs when using ONNX model with text-embeddings-inference turing image

  #544 opened Mar 27, 2025

• Could not start backend: cannot find tensor embeddings.word_embeddings.weight

  #533 opened Mar 26, 2025

• Support for mixedbread-ai/mxbai-rerank-large-v2

  #532 opened Mar 26, 2025

• error: could not compile `candle-core` (lib) due to 20 previous errors

  #528 opened Mar 24, 2025

• Relative URL without a base

  #527 opened Mar 24, 2025

• tokenize route got mismatch tokens

  #525 opened Mar 24, 2025

• Support for jina-reranker-v2-base-multilingual

  #524 opened Mar 24, 2025

• Support for Linq-AI-Research/Linq-Embed-Mistral

  #523 opened Mar 22, 2025

• Build failed due to `half` and `rand` issue

  #522 opened Mar 21, 2025

• support image embedding inference

  #521 opened Mar 21, 2025

• Support for infly/inf-retriever-v1-1.5b

  #514 opened Mar 14, 2025

• Cannot load Qodo Embed 1 1.5b (upgrade to tokenizers 0.21.0)

  #511 opened Mar 12, 2025

• Update to latest candle version?

  #508 opened Mar 6, 2025

9 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Implement the `ModernBert` model

  #459 commented on Mar 28, 2025 • 29 new comments

• Support jinaai/jina-embeddings-v3

  #418 commented on Mar 3, 2025 • 0 new comments

• support for answerdotai/ModernBERT-base

  #457 commented on Mar 3, 2025 • 0 new comments

• Using CPU Image without ONNX

  #388 commented on Mar 11, 2025 • 0 new comments

• Support NV-Embed-v2 model

  #419 commented on Mar 15, 2025 • 0 new comments

• Images Embeddings (ex. CLIP model)

  #333 commented on Mar 16, 2025 • 0 new comments

• new model lier007/xiaobu-embedding-v2

  #423 commented on Mar 25, 2025 • 0 new comments

• Move `batch`, `sort_embeddings` into `backends/candle`

  #321 commented on Mar 26, 2025 • 0 new comments

• edit logging functionality with configurable OpenTelemetry integration

  #412 commented on Mar 26, 2025 • 0 new comments