ContainersNVIDIADeepSeek-R1NVIDIALlama-3.1-Nemotron-70B-InstructNVIDIAPyTorch
CollectionsNVIDIAOmniverse Kit (FB)NVIDIADeepStream SDKNVIDIAOmniverse Kit App Streaming
ModelsNVIDIAStyleGAN3 pretrained modelsNVIDIAPeopleNetNVIDIATrafficCamNet
ResourcesNVIDIARiva Skills Quick Start
Helm ChartsNVIDIAGPU OperatorNVIDIANVIDIA NIM Operator
NVIDIA
NVIDIA
SSL En Conformer XLarge
Model
NVIDIA
NVIDIA
SSL En Conformer XLarge

Self-Supervised Learning (SSL) checkpoints for Conformer XLarge model. These are similar to w2v-Conformer model and can be fine-tuned for Automatic Speech Recognition (ASR).

1 Version
1.10.0Selected
05/26/2022 8:52 PM UTC2.37 GB
Accuracy Librispeech-100h
KeyValue
Librispeech test-other4.35 %
Librispeech dev-other4.23 %
Librispeech dev-clean2.53 %
Librispeech test-clean2.51 %
Accuracy Librispeech-960h
KeyValue
Librispeech test-other3.21 %
Librispeech dev-other3.18 %
Librispeech dev-clean1.56 %
Librispeech test-clean1.67 %
Model
KeyValue
Encoder Dimension1024
Number of Encoder Layers24
DatasetLibrilight
ARCHITECTUREConformer
INPUTS16000 KHZ MONO-CHANNEL AUDIO (WAV FILES)
Number of Weights0.6B