Simulator inference fails for a Transformer

Chih-Hsuan YenChih-Hsuan Yen
in AI Model migration

I'm trying to deploy https://huggingface.co/ainekko/smolvla_base_onnx/blob/main/smolvlm_vision.onnx, using python_api_workflow.py from exampls. With Docker on Linux, simulator inference using BIE consumes all my host memory (32GB RAM+2GB swap) and gets killed, and simulator inference using NEF returns all zeros. Is there something wrong?

Sign In or Register to comment.

© Kneron, Inc.