Relative URL without a base

[villagab4]

System Info

I am trying to run the BAAI/bge-large-en-v1.5 model

- command:
        - bash
        - -c
        - text-embeddings-router  --model-id 'BAAI/bge-large-en-v1.5' --max-batch-tokens
          10000  --max-client-batch-size 10  --payload-limit 500000000 --dtype float16

Using the image ghcr.io/huggingface/text-embeddings-inference:turing-1.5-grpc. I had updated the image to ghcr.io/huggingface/text-embeddings-inference:turing-1.6-grpc with no success.

Currently running on EKS govcloud with nvidia-smi enabled. Everything has been working for 6 months with no changes to the deployment. Now, I am getting these errors

kubectl logs goveagle-ai-services-embedding-c4c5966b9-pjmlc
2025-03-24T20:27:59.173912Z  INFO text_embeddings_router: router/src/main.rs:175: Args { model_id: "BAA*/***-*****-**-v1.5", revision: None, tokenization_workers: None, dtype: Some(Float16), pooling: None, max_concurrent_requests: 512, max_batch_tokens: 10000, max_batch_requests: None, max_client_batch_size: 10, auto_truncate: false, default_prompt_name: None, default_prompt: None, hf_api_token: None, hostname: "goveagle-ai-services-embedding-c4c5966b9-pjmlc", port: 80, uds_path: "/tmp/text-embeddings-inference-server", huggingface_hub_cache: Some("/data"), payload_limit: 500000000, api_key: None, json_output: false, otlp_endpoint: None, otlp_service_name: "text-embeddings-inference.server", cors_allow_origin: None }
2025-03-24T20:27:59.174025Z  INFO hf_hub: /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/hf-hub-0.3.2/src/lib.rs:55: Token file not found "/root/.cache/huggingface/token"    
2025-03-24T20:27:59.262552Z  INFO download_artifacts: text_embeddings_core::download: core/src/download.rs:20: Starting download
2025-03-24T20:27:59.262574Z  INFO download_artifacts:download_pool_config: text_embeddings_core::download: core/src/download.rs:53: Downloading `1_Pooling/config.json`
2025-03-24T20:27:59.334606Z  WARN download_artifacts: text_embeddings_core::download: core/src/download.rs:26: Download failed: request error: builder error: relative URL without a base
2025-03-24T20:27:59.577727Z  INFO download_artifacts:download_new_st_config: text_embeddings_core::download: core/src/download.rs:77: Downloading `config_sentence_transformers.json`
2025-03-24T20:27:59.608809Z  WARN download_artifacts: text_embeddings_core::download: core/src/download.rs:36: Download failed: request error: builder error: relative URL without a base
2025-03-24T20:27:59.608832Z  INFO download_artifacts: text_embeddings_core::download: core/src/download.rs:40: Downloading `config.json`
Error: Could not download model artifacts

Caused by:
    0: request error: builder error: relative URL without a base
    1: builder error: relative URL without a base
    2: relative URL without a base

Information

• Docker
• The CLI directly

Tasks

• An officially supported command
• My own modifications

Reproduction

Deploy a helm chart with the commands listed above

Expected behavior

I am expecting the model to download properly

[Narsil]

Can you try with 1.6.0 to check ? We have been upgrading dependencies quite heavily recently, it's possible the version you had included an issue, even though I cannot reproduce on my end.

Do you use any special environment flags besides the command line ?

We're going to release 1.7.0 soon to include all those dependency upgrades and fixes, it shouldn't contain any bug anymore.

[Narsil]

1.7.0 is out, and works perfectly on our end, and even with 1.6.1 we're not able to reproduce the issue.

[vrdn-23]

The issue might be related to specifying the hostname hostname: "goveagle-ai-services-embedding-c4c5966b9-pjmlc" without an http/https schema.

[villagab4]

This problem resurfaced today. Upgrading to 1.7 fixed the issue

[rjrobben]

When will 1.7 be pushed to the Google Artifact Registry?

Encountered same error while deploying using cloud run with the latest:
gcr.io/deeplearning-platform-release/huggingface-text-embeddings-inference-cpu.1-6

Many thanks.

[chensuyue]

I got the same issue with cpu-1.5. opea-project/GenAIExamples#2098