NeMo
NeMo copied to clipboard
NVIDIA
A scalable generative AI framework built for researchers and developers working on Large Language Models, Multimodal, and Speech AI (Automatic Speech Recognition and Text-to-Speech)
**Describe the bug** Hi team, I was experimenting with streaming sortformer and discovered that it was not picking up `duration` parameter from the input manifest. Based on HF instructions the...
**Is your feature request related to a problem? Please describe.** Based on the problem described here: https://github.com/NVIDIA-NeMo/NeMo/issues/15143 The "nvidia/parakeet-tdt-0.6b-v3" model sometimes returns different amounts of transcribed words and inferred word...
**Describe the bug** I am trying to follow method header of the transcribe method that claims that you can give the models a numpy array of audio. This works for...
## Describe the bug Running RNNT ASR inference with NeMo 2.6.0 on an NVIDIA A100 (CUDA 12.8, PyTorch 2.9.1+cu128) fails during decoding due to a CUDA Graphs initialization error: ```text...
Dear Team, examples/asr/asr_hybrid_transducer_ctc/speech_to_text_hybrid_rnnt_ctc_bpe.py is not working with numba-cuda>0.15.1. Only after installation of numba-cuda with less version it works. ` pip install numba-cuda==0.15.1 ` Please list *minimal* steps or code snippet...
**Describe the bug** I am trying to do inference on a finetuned `nvidia/parakeet-tdt-0.6b-v3` model. We have an efficient inference framework that feeds in audio batches (batch-size=8) into the model without...
Hi, I noticed that `nemo-toolkit[asr]` currently restricts the `numpy` dependency to
> [!IMPORTANT] > The `Update branch` button must only be pressed in very rare occassions. > An outdated branch is never blocking the merge of a PR. > Please reach...
Hi there, i was looking for a method to perform [context biasing](https://github.com/NVIDIA-NeMo/NeMo/pull/8223) based on the [example notebook](https://github.com/NVIDIA-NeMo/NeMo/blob/main/tutorials/asr/ASR_Context_Biasing.ipynb) on the Parakeet tdt 0.6b v3 model. I added the EncDecHybridRNNTCTCModel but got...
**Describe the bug** Loss is always NaN (See image below). In nemo 2.4 it works fine : Could be a problem with a package or something? **Steps/Code to reproduce bug**...
