lorax
lorax copied to clipboard
Kernel issues, with docker image:main
System Info
spec - aws g6e.12xLarge
Hi, I'm trying out lorax. I ran a docker container with image tag as main(ghcr.io/predibase/lorax:main) and was facing some kernel issues. Attaching logs. After changing the image tag to latest, the server has started. Reporting this issue, so it can help you debug and fix. Thank you
indicies, layer_idx, 1.0)\nRuntimeError: No suitable kernel. h_in=256 h_out=1024 dtype=BFloat16\n"},"target":"lorax_launcher"}
{"timestamp":"2025-04-02T05:56:29.502899Z","level":"ERROR","message":"Server error: No suitable kernel. h_in=256 h_out=1024 dtype=BFloat16","target":"lorax_client","filename":"router/client/src/lib.rs","line_number":38,"span":{"name":"warmup"},"spans":[{"max_input_length":4095,"max_prefill_tokens":4145,"max_total_tokens":4096,"name":"warmup"},{"name":"warmup"}]}
{"timestamp":"2025-04-02T05:56:29.504029Z","level":"ERROR","message":"Server error: No suitable kernel. h_in=256 h_out=1024 dtype=BFloat16","target":"lorax_client","filename":"router/client/src/lib.rs","line_number":38,"span":{"name":"warmup"},"spans":[{"max_input_length":4095,"max_prefill_tokens":4145,"max_total_tokens":4096,"name":"warmup"},{"name":"warmup"}]}
{"timestamp":"2025-04-02T05:56:29.504301Z","level":"ERROR","message":"Server error: No suitable kernel. h_in=256 h_out=1024 dtype=BFloat16","target":"lorax_client","filename":"router/client/src/lib.rs","line_number":38,"span":{"name":"warmup"},"spans":[{"max_input_length":4095,"max_prefill_tokens":4145,"max_total_tokens":4096,"name":"warmup"},{"name":"warmup"}]}
{"timestamp":"2025-04-02T05:56:29.506594Z","level":"ERROR","message":"Server error: No suitable kernel. h_in=256 h_out=1024 dtype=BFloat16","target":"lorax_client","filename":"router/client/src/lib.rs","line_number":38,"span":{"name":"warmup"},"spans":[{"max_input_length":4095,"max_prefill_tokens":4145,"max_total_tokens":4096,"name":"warmup"},{"name":"warmup"}]}
Error: Warmup(Generation("No suitable kernel. h_in=256 h_out=1024 dtype=BFloat16"))
Information
- [x] Docker
- [ ] The CLI directly
Tasks
- [x] An officially supported command
- [ ] My own modifications
Reproduction
FROM ghcr.io/predibase/lorax:main
ENV HUGGINGFACE_HUB_CACHE=/data
ENV HF_HUB_ENABLE_HF_TRANSFER=1
ENTRYPOINT ["lorax-launcher", "--json-output", "--model-id", "meta-llama/Llama-3.1-70B-Instruct", "--num-shard", "4", "--port", "80"]
docker build -f Dockerfile . -t lorax volume=$PWD/data docker run --gpus all --env-file .env --shm-size 1g -p 8080:80 -v $volume:/data lorax
Expected behavior
The webserver should start