AmarnathVoicing comments

Results 5 comments of


                                            AmarnathVoicing

GPU memory is not released by Triton

I have an issue with memory allocation/deallocation with this module in the triton server, ` name: "decoder" backend: "onnxruntime" max_batch_size: 0 input [ { name: "ASR", data_type: TYPE_FP32, dims: [...

GPU memory is not released by Triton

Thank you! Will try this.

Poor WER when trying to fine-tune Parakeet v2 TDT to other dataset than English

Hello @jeremy110, I am trying to train 0.6B model(en) with 900hrs of training data (eng+hindi). Is it recommended to train encoder as well or decoder is enough?

Poor WER when trying to fine-tune Parakeet v2 TDT to other dataset than English

Thank you. I will use the third approach as i am using code switching data and can use aggregate tokenizer. The 3rd point is really helpful.

Optimize Dynamic Shape Inference for TTS Model with HiFi-GAN Vocoder

@UmerrAhsan Did you solve for memory allocation/deallocation in this? If so, can you share the method? I am getting high latency of more than 1s because of this.