Support jinaai/jina-embeddings-v3

[luonist]

Model description

jina-embeddings-v3 is a multilingual multi-task text embedding model designed for a variety of NLP applications. Based on the Jina-XLM-RoBERTa architecture, this model supports Rotary Position Embeddings to handle long input sequences up to 8192 tokens. Additionally, it features 5 LoRA adapters to generate task-specific embeddings efficiently.

Open source status

• The model implementation is available
• The model weights are available

Provide useful links for the implementation

No response

[trillionmonster]

wait for Support News

[ilhamdwiprastyo]

Hi everyone,

I'm encountering an issue while trying to run text-embeddings-inference with Jina embeddings v3 model. The error suggests there's a problem with the model configuration:

Error:
Caused by:
missing field model_type at line 51 column 1
Error: Failed to parse config.json

[dongfangduoshou123]

wait for Support News

[allisoneer]

Heyo! I've been subscribed to this issue for a little while now and keep seeing the comment about 'Support News' - I'm new to this repository and the HF github ecosystem and am wondering what exactly this is referring to? Is there a specific place where support news is released/announced? Just trying to ensure I track the updates in the right place: Thanks!

[19920716]

expectation this version

[RRaphaell]

any updates on this?
it shows this error for me:

2025-02-19T10:18:04.827567Z ERROR text_embeddings_backend: backends/src/lib.rs:381: Could not start Candle backend: Could not start backend: Model is not supported

Caused by:
    unknown variant `rotary`, expected one of `absolute`, `alibi`, `rope`
Error: Could not create backend

Caused by:
    Could not start backend: Could not start a suitable backend

[dromeuf]

same problem.

(HF_TEI) dromeuf@MAIA:~$ sudo docker run --gpus all -p 33438:80 ghcr.io/huggingface/text-embeddings-inference:latest --model-id jinaai/jina-embeddings-v3
                       
2025-02-27T16:14:44.416062Z  INFO text_embeddings_router: router/src/main.rs:175: Args { model_id: "jin***/****-*********s-v3", revision: None, tokenization_workers: None, dtyp
e: None, pooling: None, max_concurrent_requests: 512, max_batch_tokens: 16384, max_batch_requests: None, max_client_batch_size: 32, auto_truncate: false, default_prompt_name: N
one, default_prompt: None, hf_api_token: None, hostname: "d5a45fdddef2", port: 80, uds_path: "/tmp/text-embeddings-inference-server", huggingface_hub_cache: Some("/data"), payl
oad_limit: 2000000, api_key: None, json_output: false, otlp_endpoint: None, otlp_service_name: "text-embeddings-inference.server", cors_allow_origin: None }                    
2025-02-27T16:14:44.416173Z  INFO hf_hub: /root/.cargo/registry/src/index.crates.io-1949cf8c6b5b557f/hf-hub-0.3.2/src/lib.rs:55: Token file not found "/root/.cache/huggingface/
token"                                                                                                                                                                          
2025-02-27T16:14:44.483283Z  INFO download_artifacts: text_embeddings_core::download: core/src/download.rs:20: Starting download                                                
2025-02-27T16:14:44.483310Z  INFO download_artifacts:download_pool_config: text_embeddings_core::download: core/src/download.rs:53: Downloading `1_Pooling/config.json`         
2025-02-27T16:14:45.513500Z  INFO download_artifacts:download_new_st_config: text_embeddings_core::download: core/src/download.rs:77: Downloading `config_sentence_transformers.
json`                                                                                                                                                                           
2025-02-27T16:14:45.723275Z  INFO download_artifacts: text_embeddings_core::download: core/src/download.rs:40: Downloading `config.json`                                        
2025-02-27T16:14:46.242315Z  INFO download_artifacts: text_embeddings_core::download: core/src/download.rs:43: Downloading `tokenizer.json`                                     
2025-02-27T16:14:48.200382Z  INFO download_artifacts: text_embeddings_core::download: core/src/download.rs:47: Model artifacts downloaded in 3.717100569s                       
Error: Failed to parse `config.json`                                                                                                                                            
                                                                                                                                                                                
Caused by:                                                                                                                                                                      
    missing field `model_type` at line 51 column 1                                                                                                                              
                                                                                                                                                      

Thanks for your great work, Kind regards, David.

[pocman]

jinaai/jina-embeddings-v3 model is exactly what I'm looking for as an alternative to GCP gecko endpoint for classification. It would be awesome to have it supported in TEI !

The classification adapter generates embeddings that are effective for training downstream classification models, particularly logistic regression classifiers. To train the adapter, we employ the classification training method proposed for the Gecko embedding model [Lee et al., 2024a]. Specifically, we select datasets covering a range of common classification tasks, including sentiment analysis, intent classification, and article categorization.
Source

[dromeuf]

It work with OpenWebUI SentenceTransformer, but without flash support on my pc, which takes a very long time to vectorize...