Qwen3-Reranker-0.6B start error

[Shengshenlan]

model=/root/tei/data/Qwen3-Reranker-0.6B

./target/release/text-embeddings-router --model-id $model --port 8023 --auto-truncate

error

(ms-swift) root@u:~/text-embeddings-inference# bash test.sh 
2025-11-24T13:41:47.443474Z  INFO text_embeddings_router: router/src/main.rs:203: Args { model_id: "/roo*/***/****/*****-********-0.6B", revision: None, tokenization_workers: None, dtype: None, pooling: None, max_concurrent_requests: 512, max_batch_tokens: 16384, max_batch_requests: None, max_client_batch_size: 32, auto_truncate: true, default_prompt_name: None, default_prompt: None, dense_path: None, hf_api_token: None, hf_token: None, hostname: "0.0.0.0", port: 8023, uds_path: "/tmp/text-embeddings-inference-server", huggingface_hub_cache: None, payload_limit: 2000000, api_key: None, json_output: false, disable_spans: false, otlp_endpoint: None, otlp_service_name: "text-embeddings-inference.server", prometheus_port: 9000, cors_allow_origin: None }
2025-11-24T13:41:47.645381Z  WARN text_embeddings_router: router/src/lib.rs:191: Could not find a Sentence Transformers config
2025-11-24T13:41:47.645397Z  WARN text_embeddings_router: router/src/lib.rs:205: The input sequences will be truncated to 16384 tokens even if the model `max_input_length` is greater than the provided `--max-batch-tokens` (40960 > 16384), as `--auto-truncate` is enabled.
2025-11-24T13:41:47.645401Z  INFO text_embeddings_router: router/src/lib.rs:216: Maximum number of tokens per request: 16384
2025-11-24T13:41:47.645515Z  INFO text_embeddings_core::tokenization: core/src/tokenization.rs:38: Starting 24 tokenization workers
2025-11-24T13:41:48.095787Z  INFO text_embeddings_router: router/src/lib.rs:264: Starting model backend
2025-11-24T13:41:48.097062Z  INFO text_embeddings_backend_candle: backends/candle/src/lib.rs:305: Starting Qwen3 model on Cpu
2025-11-24T13:41:48.097203Z ERROR text_embeddings_backend: backends/src/lib.rs:481: Could not start Candle backend: Could not start backend: `classifier` model type is not supported for Qwen3
Error: Could not create backend

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

this local install text-embeddings-router

I used lastest package

(ms-swift) root@u:~/text-embeddings-inference# git log -1 --oneline
106d25f (HEAD -> main, origin/main, origin/HEAD) Fix `TruncationDirection` to deserialize from lowercase and capitalized (#755)

I have read #698

and

I have used convert_to_st.py

What should I do then next?

[alvarobartt]

Hey @Shengshenlan, thanks for reporting! This error is kind of expected given that Qwen3-Reranker is not yet implemented but there are some PRs on the way and I'm planning to merge that for v1.9.0, in the meantime the error that you're seeing is expected, apologies for the inconvenience! 🤗