Elias

Still relevant for https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/stt_ru_fastconformer_hybrid_large_pc docker: nvcr.io/nvidia/nemo:24.01.speech And this model has a different tokenizer from the one in ngc. https://huggingface.co/nvidia/stt_ru_fastconformer_hybrid_large_pc/tree/main

There is a solution for a single file. Take the `.wav` file, rename it to the name of our input. For example `IN` for config above. And put it in...

@matthewkotila, It's not about multiple inputs. It's about multiple requests. With the --input-data parameter, I can only send 1 file per input from the `data/` folder. But I want to...

@matthewkotila If I use b64 + json, I will need to change the logic of the triton service, right? Would need to decode b64.

Full debug log: [debug_tensorrt_conformer_transducer.log](https://github.com/pytorch/TensorRT/files/12342272/debug_tensorrt_conformer_transducer.log)

updated the code for full reproduction^ P.S. no one conformer model from NeMo is compiled to tensorrt I tried stt_en_conformer_transducer_large, stt_en_conformer_transducer_small

Thanks! Do I understand correctly that if I try to compile this model from native pytorch to tensorrt it might work? Or is the problem in the Conformer architecture itself?

@gs-olive hello! I tried to use your code, but I got an error `torch._export.verifier.SpecViolationError: Node.meta reshape_default is missing val field.` log: [conf_trt_log.log](https://github.com/pytorch/TensorRT/files/14355033/conf_trt_log.log) debug log: [conf_trt_debug_log.log](https://github.com/pytorch/TensorRT/files/14355283/conf_trt_debug_log.log) image: nvcr.io/nvidia/tensorrt:23.07-py3 python: 3.10.6 reqs:...

It works, thanks! But the Nemo model weighs 56Mb, after conversion the weight of the trt becomes ~800Mb for input tensor [1, 80, 6000]. It's ok? `-rw-r--r-- 1 root root...

[conf_trt_ts_debug.log](https://github.com/pytorch/TensorRT/files/14424948/conf_trt_ts_debug.log) Is that enough? If you export this model to torchscript or onnx, it decomposes into two files: encoder.ts and decoder.ts Maybe that's why there are some problems in converting...