SiCong Li comments

Repositories
Issues
Comments

Results 2 comments of


                                            SiCong Li

Segmentation fault running inference on gemma3

Hi @HarishPermal thanks for the reply! We found out that after renaming the signatures and input/output tensors for the gemma3 model, the segfault went away: ``` -auto* prefill_runner = llm->interpreter->GetSignatureRunner("prefill");...

Invalid .tflite Flatbuffer issue for llm inference task on Android CPU

Hi @akashvverma1995 Thanks for your reply! However we are trying to use directly the C API defined in mediapipe/tasks/cc/genai/inference for a native app, instead of using it as an Android....