Downloads · 30 days
0
apbaxel/sherpa-onnx-nemotron-3.5-asr-streaming-0.6b-int8
sherpa-onnx-nemotron-3.5-asr-streaming-0.6b-int8 is a machine learning model from apbaxel. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for sherpa-onnx. The card lists the license as other.
Mirror of the k2-fsa/sherpa-onnx ASR-models release package sherpa-onnx-nemotron-3.5-asr-streaming-0.6b-560ms-int8-2026-06-11 — the published Nemotron 3.5 ASR Streaming 0.6B streaming transducer package, int8-quantize…
Downloads · 30 days
0
Access
Public
Updated Aug 8, 2026
Repo size
682 MB
Likes
0
Public
Click a slice to open those files.
.onnx682 MB · 100%
From the Hugging Face model README
Mirror of the k2-fsa/sherpa-onnx ASR-models release package sherpa-onnx-nemotron-3.5-asr-streaming-0.6b-560ms-int8-2026-06-11 — the published Nemotron 3.5 ASR Streaming 0.6B streaming transducer package, int8-quantized, trained for the 560ms chunk size. Packaged by k2-fsa against sherpa-onnx v1.13.4.
| File | Size | Purpose |
|---|---|---|
encoder.int8.onnx | 657,601,403 bytes | Streaming RNN-T encoder |
decoder.int8.onnx | 14,978,075 bytes | Streaming RNN-T decoder |
joiner.int8.onnx | 9,504,438 bytes | Streaming RNN-T joint network |
tokens.txt | 131,440 bytes | BPE token vocabulary (13088 entries + blank) |
This mirror redistributes the model weights exactly as packaged upstream — no re-quantization, no re-export, no modification. The four files above are byte-for-byte identical with the original package contents (sha256-verified at upload).
These model weights are licensed under the NVIDIA Open Model Data Warehouse License Agreement v1.1 (OpenMDW-1.1).
Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
NVIDIA Corp. ("NVIDIA") agrees to provide, and You agree to receive and use, the Models, Derivative Models, and Related Materials (collectively, "Materials") solely under the terms and conditions of this Open Model Data Warehouse License Agreement v1.1 ("Agreement").
1. DEFINITIONS
1.1. "Models" means the NVIDIA thirty-five (35) billion parameter language model known as Nemotron, including the family of dense and Minitron models, and any derivatives made by NVIDIA, including fine-tunes, instruction tuned varieties, and adapters, all made available under this OpenMDW License Agreement v1.1.
1.2. "Derivative Models" means any model created by You from a Model or output or result of a Model, including by fine-tuning, pruning, distilling, retraining, quantizing, re-quantizing, or otherwise modifying a Model, its outputs, or its results.
1.3. "Related Materials" means the documentation, source code, sample code, and data (excluding Models) made available by NVIDIA under this Agreement.
1.4. "You" or "Your" means the individual or entity entering into this Agreement with NVIDIA.
2. LICENSE GRANT
2.1. Subject to Your compliance with the terms and conditions of this Agreement, NVIDIA hereby grants You a non-exclusive, worldwide, royalty-free, non-sublicensable license to:
(a) use, reproduce, modify, and distribute the Models and Derivative Models; and
(b) use, reproduce, modify, and distribute the Related Materials.
3. REDISTRIBUTION
3.1. You may distribute copies of the Models, Derivative Models, and Related Materials under the terms of this Agreement, provided:
(a) You make available a copy of this Agreement to recipients; and
(b) You reproduce all copyright, trademark, and other proprietary notices contained in the Materials in all copies of the Materials You distribute.
3.2. Any modification or Derivative Model You distribute must be marked or labeled in a reasonable manner to indicate the modification made.
4. ATTRIBUTION NOTICE
The following attribution notice must accompany the Models and Derivative Models:
"Nemotron" model family © NVIDIA CORPORATION & AFFILIATES.
Licensed under the NVIDIA Open Model Data Warehouse License Agreement v1.1 (OpenMDW-1.1).
5. NO WARRANTIES + LIMITATION OF LIABILITY
THE MATERIALS ARE PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND. NVIDIA DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NONINFRINGEMENT. IN NO EVENT SHALL NVIDIA BE LIABLE FOR ANY SPECIAL, INCIDENTAL, INDIRECT, OR CONSEQUENTIAL DAMAGES ARISING OUT OF OR RELATED TO THIS AGREEMENT OR THE MATERIALS.
6. TRADEMARKS
No trademark licenses are granted under this Agreement. "NVIDIA" and the NVIDIA logo are trademarks of NVIDIA Corporation. Nothing in this Agreement grants You any rights to use NVIDIA's trade names, trademarks, or other brand identifiers except as required to reproduce the attribution notice in §4 above.
7. TERMINATION
NVIDIA may terminate this Agreement with You if You breach any of its terms. Upon termination, You must cease using the Materials and destroy all copies in Your possession or control.
8. UPSTREAM COMPLIANCE
This mirror is provided as-is for downstream consumers. The redistributor (apbaxel) is in compliance with OpenMDW-1.1 §3 (Redistribution) by making this Agreement available to all recipients + reproducing the copyright, trademark, and proprietary notices contained in the Materials. Upstream OpenMDW-1.1 text is published by NVIDIA at https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b/blob/main/OpenMDW-1.1.txt.
If you use sherpa-onnx in your research, please cite the sherpa-onnx project (https://github.com/k2-fsa/sherpa-onnx).