mirror of nvidia/canary-qwen-2.5b

.eval_results/open_asr_leaderboard.yaml

@@ -0,0 +1,99 @@
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: mean_wer
+  value: 5.63
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio
+
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: rtfx
+  value: 418.28
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio
+
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: ami_wer
+  value: 10.19
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio
+
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: earnings22_wer
+  value: 10.45
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio
+
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: gigaspeech_wer
+  value: 9.43
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio
+
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: librispeech_clean_wer
+  value: 1.61
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio
+
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: librispeech_other_wer
+  value: 3.1
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio
+
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: spgispeech_wer
+  value: 1.9
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio
+
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: tedlium_wer
+  value: 2.71
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio
+
+- dataset:
+    id: hf-audio/open-asr-leaderboard
+    task_id: voxpopuli_wer
+  value: 5.66
+  date: '2025-06-26'
+  source:
+    url: https://huggingface.co/hf-audio
+    name: open-asr-leaderboard
+    user: hf-audio

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+results_hf/open-asr-leaderboarddatasets-test-only-spgispeech-test.jsonl filter=lfs diff=lfs merge=lfs -text

LICENSES

@@ -0,0 +1,103 @@
+# Copyright NVIDIA "Canary-1B-Flash" - CC-BY-4.0 License
+License Text ( https://spdx.org/licenses/CC-BY-4.0.html )
+Creative Commons Attribution 4.0 International Public License
+
+By exercising the Licensed Rights (defined below), You accept and agree to be bound by the terms and conditions of this Creative Commons Attribution 4.0 International Public License ("Public License"). To the extent this Public License may be interpreted as a contract, You are granted the Licensed Rights in consideration of Your acceptance of these terms and conditions, and the Licensor grants You such rights in consideration of benefits the Licensor receives from making the Licensed Material available under these terms and conditions.
+
+Section 1 Definitions.
+
+a. Adapted Material means material subject to Copyright and Similar Rights that is derived from or based upon the Licensed Material and in which the Licensed Material is translated, altered, arranged, transformed, or otherwise modified in a manner requiring permission under the Copyright and Similar Rights held by the Licensor. For purposes of this Public License, where the Licensed Material is a musical work, performance, or sound recording, Adapted Material is always produced where the Licensed Material is synched in timed relation with a moving image.
+b. Adapter's License means the license You apply to Your Copyright and Similar Rights in Your contributions to Adapted Material in accordance with the terms and conditions of this Public License.
+c. Copyright and Similar Rights means copyright and/or similar rights closely related to copyright including, without limitation, performance, broadcast, sound recording, and Sui Generis Database Rights, without regard to how the rights are labeled or categorized. For purposes of this Public License, the rights specified in Section 2(b)(1)-(2) are not Copyright and Similar Rights.
+d. Effective Technological Measures means those measures that, in the absence of proper authority, may not be circumvented under laws fulfilling obligations under Article 11 of the WIPO Copyright Treaty adopted on December 20, 1996, and/or similar international agreements.
+e. Exceptions and Limitations means fair use, fair dealing, and/or any other exception or limitation to Copyright and Similar Rights that applies to Your use of the Licensed Material.
+f. Licensed Material means the artistic or literary work, database, or other material to which the Licensor applied this Public License.
+g. Licensed Rights means the rights granted to You subject to the terms and conditions of this Public License, which are limited to all Copyright and Similar Rights that apply to Your use of the Licensed Material and that the Licensor has authority to license.
+h. Licensor means the individual(s) or entity(ies) granting rights under this Public License.
+i. Share means to provide material to the public by any means or process that requires permission under the Licensed Rights, such as reproduction, public display, public performance, distribution, dissemination, communication, or importation, and to make material available to the public including in ways that members of the public may access the material from a place and at a time individually chosen by them.
+j. Sui Generis Database Rights means rights other than copyright resulting from Directive 96/9/EC of the European Parliament and of the Council of 11 March 1996 on the legal protection of databases, as amended and/or succeeded, as well as other essentially equivalent rights anywhere in the world.
+k. You means the individual or entity exercising the Licensed Rights under this Public License. Your has a corresponding meaning.
+Section 2 Scope.
+
+a. License grant.
+1. Subject to the terms and conditions of this Public License, the Licensor hereby grants You a worldwide, royalty-free, non-sublicensable, non-exclusive, irrevocable license to exercise the Licensed Rights in the Licensed Material to:
+A. reproduce and Share the Licensed Material, in whole or in part; and
+B. produce, reproduce, and Share Adapted Material.
+2. Exceptions and Limitations. For the avoidance of doubt, where Exceptions and Limitations apply to Your use, this Public License does not apply, and You do not need to comply with its terms and conditions.
+3. Term. The term of this Public License is specified in Section 6(a).
+4. Media and formats; technical modifications allowed. The Licensor authorizes You to exercise the Licensed Rights in all media and formats whether now known or hereafter created, and to make technical modifications necessary to do so. The Licensor waives and/or agrees not to assert any right or authority to forbid You from making technical modifications necessary to exercise the Licensed Rights, including technical modifications necessary to circumvent Effective Technological Measures. For purposes of this Public License, simply making modifications authorized by this Section 2(a)(4) never produces Adapted Material.
+5. Downstream recipients.
+A. Offer from the Licensor – Licensed Material. Every recipient of the Licensed Material automatically receives an offer from the Licensor to exercise the Licensed Rights under the terms and conditions of this Public License.
+B. No downstream restrictions. You may not offer or impose any additional or different terms or conditions on, or apply any Effective Technological Measures to, the Licensed Material if doing so restricts exercise of the Licensed Rights by any recipient of the Licensed Material.
+6. No endorsement. Nothing in this Public License constitutes or may be construed as permission to assert or imply that You are, or that Your use of the Licensed Material is, connected with, or sponsored, endorsed, or granted official status by, the Licensor or others designated to receive attribution as provided in Section 3(a)(1)(A)(i).
+b. Other rights.
+1. Moral rights, such as the right of integrity, are not licensed under this Public License, nor are publicity, privacy, and/or other similar personality rights; however, to the extent possible, the Licensor waives and/or agrees not to assert any such rights held by the Licensor to the limited extent necessary to allow You to exercise the Licensed Rights, but not otherwise.
+2. Patent and trademark rights are not licensed under this Public License.
+3. To the extent possible, the Licensor waives any right to collect royalties from You for the exercise of the Licensed Rights, whether directly or through a collecting society under any voluntary or waivable statutory or compulsory licensing scheme. In all other cases the Licensor expressly reserves any right to collect such royalties.
+Section 3 License Conditions.
+
+Your exercise of the Licensed Rights is expressly made subject to the following conditions.
+
+a. Attribution.
+1. If You Share the Licensed Material (including in modified form), You must:
+A. retain the following if it is supplied by the Licensor with the Licensed Material:
+i. identification of the creator(s) of the Licensed Material and any others designated to receive attribution, in any reasonable manner requested by the Licensor (including by pseudonym if designated);
+ii. a copyright notice;
+iii. a notice that refers to this Public License;
+iv. a notice that refers to the disclaimer of warranties;
+v. a URI or hyperlink to the Licensed Material to the extent reasonably practicable;
+B. indicate if You modified the Licensed Material and retain an indication of any previous modifications; and
+C. indicate the Licensed Material is licensed under this Public License, and include the text of, or the URI or hyperlink to, this Public License.
+2. You may satisfy the conditions in Section 3(a)(1) in any reasonable manner based on the medium, means, and context in which You Share the Licensed Material. For example, it may be reasonable to satisfy the conditions by providing a URI or hyperlink to a resource that includes the required information.
+3. If requested by the Licensor, You must remove any of the information required by Section 3(a)(1)(A) to the extent reasonably practicable.
+4. If You Share Adapted Material You produce, the Adapter's License You apply must not prevent recipients of the Adapted Material from complying with this Public License.
+Section 4 Sui Generis Database Rights.
+
+Where the Licensed Rights include Sui Generis Database Rights that apply to Your use of the Licensed Material:
+
+a. for the avoidance of doubt, Section 2(a)(1) grants You the right to extract, reuse, reproduce, and Share all or a substantial portion of the contents of the database;
+b. if You include all or a substantial portion of the database contents in a database in which You have Sui Generis Database Rights, then the database in which You have Sui Generis Database Rights (but not its individual contents) is Adapted Material; and
+c. You must comply with the conditions in Section 3(a) if You Share all or a substantial portion of the contents of the database.
+For the avoidance of doubt, this Section 4 supplements and does not replace Your obligations under this Public License where the Licensed Rights include other Copyright and Similar Rights.
+
+Section 5 Disclaimer of Warranties and Limitation of Liability.
+
+a. Unless otherwise separately undertaken by the Licensor, to the extent possible, the Licensor offers the Licensed Material as-is and as-available, and makes no representations or warranties of any kind concerning the Licensed Material, whether express, implied, statutory, or other. This includes, without limitation, warranties of title, merchantability, fitness for a particular purpose, non-infringement, absence of latent or other defects, accuracy, or the presence or absence of errors, whether or not known or discoverable. Where disclaimers of warranties are not allowed in full or in part, this disclaimer may not apply to You.
+b. To the extent possible, in no event will the Licensor be liable to You on any legal theory (including, without limitation, negligence) or otherwise for any direct, special, indirect, incidental, consequential, punitive, exemplary, or other losses, costs, expenses, or damages arising out of this Public License or use of the Licensed Material, even if the Licensor has been advised of the possibility of such losses, costs, expenses, or damages. Where a limitation of liability is not allowed in full or in part, this limitation may not apply to You.
+c. The disclaimer of warranties and limitation of liability provided above shall be interpreted in a manner that, to the extent possible, most closely approximates an absolute disclaimer and waiver of all liability.
+Section 6 Term and Termination.
+
+a. This Public License applies for the term of the Copyright and Similar Rights licensed here. However, if You fail to comply with this Public License, then Your rights under this Public License terminate automatically.
+b. Where Your right to use the Licensed Material has terminated under Section 6(a), it reinstates:
+1. automatically as of the date the violation is cured, provided it is cured within 30 days of Your discovery of the violation; or
+2. upon express reinstatement by the Licensor.
+c. For the avoidance of doubt, this Section 6(b) does not affect any right the Licensor may have to seek remedies for Your violations of this Public License.
+d. For the avoidance of doubt, the Licensor may also offer the Licensed Material under separate terms or conditions or stop distributing the Licensed Material at any time; however, doing so will not terminate this Public License.
+e. Sections 1, 5, 6, 7, and 8 survive termination of this Public License.
+Section 7 Other Terms and Conditions.
+
+a. The Licensor shall not be bound by any additional or different terms or conditions communicated by You unless expressly agreed.
+b. Any arrangements, understandings, or agreements regarding the Licensed Material not stated herein are separate from and independent of the terms and conditions of this Public License.
+Section 8 Interpretation.
+
+a. For the avoidance of doubt, this Public License does not, and shall not be interpreted to, reduce, limit, restrict, or impose conditions on any use of the Licensed Material that could lawfully be made without permission under this Public License.
+b. To the extent possible, if any provision of this Public License is deemed unenforceable, it shall be automatically reformed to the minimum extent necessary to make it enforceable. If the provision cannot be reformed, it shall be severed from this Public License without affecting the enforceability of the remaining terms and conditions.
+c. No term or condition of this Public License will be waived and no failure to comply consented to unless expressly agreed to by the Licensor.
+d. Nothing in this Public License constitutes or may be interpreted as a limitation upon, or waiver of, any privileges and immunities that apply to the Licensor or You, including from the legal processes of any jurisdiction or authority.
+
+# Copyright Alibaba Cloud - "Qwen3-1.7B" - Apache 2.0 License
+License Text ( https://spdx.org/licenses/Apache-2.0.html )
+
+Copyright 2025 Alibaba Cloud
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.

README.md

@@ -0,0 +1,485 @@
+---
+license: cc-by-4.0
+language:
+- en
+library_name: nemo
+datasets:
+- Granary
+- YTC
+- Yodas2
+- LibriLight
+- librispeech_asr
+- fisher_corpus
+- Switchboard-1
+- WSJ-0
+- WSJ-1
+- National-Singapore-Corpus-Part-1
+- National-Singapore-Corpus-Part-6
+- vctk
+- voxpopuli
+- europarl
+- multilingual_librispeech
+- fleurs
+- mozilla-foundation/common_voice_8_0
+- MLCommons/peoples_speech
+thumbnail: null
+tags:
+- automatic-speech-recognition
+- speech
+- audio
+- Transformer
+- FastConformer
+- Conformer
+- pytorch
+- NeMo
+- Qwen
+- hf-asr-leaderboard
+widget:
+- example_title: Librispeech sample 1
+  src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
+- example_title: Librispeech sample 2
+  src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
+model-index:
+- name: canary-qwen-2.5b
+  results:
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: AMI (Meetings test)
+      type: edinburghcstr/ami
+      config: ihm
+      split: test
+      args:
+        language: en
+    metrics:
+    - name: Test WER
+      type: wer
+      value: 10.19
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: Earnings-22
+      type: revdotcom/earnings22
+      split: test
+      args:
+        language: en
+    metrics:
+    - name: Test WER
+      type: wer
+      value: 10.45
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: GigaSpeech
+      type: speechcolab/gigaspeech
+      split: test
+      args:
+        language: en
+    metrics:
+    - name: Test WER
+      type: wer
+      value: 9.43
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: LibriSpeech (clean)
+      type: librispeech_asr
+      config: other
+      split: test
+      args:
+        language: en
+    metrics:
+    - name: Test WER
+      type: wer
+      value: 1.61
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: LibriSpeech (other)
+      type: librispeech_asr
+      config: other
+      split: test
+      args:
+        language: en
+    metrics:
+    - name: Test WER
+      type: wer
+      value: 3.1
+  - task:
+      type: Automatic Speech Recognition
+      name: automatic-speech-recognition
+    dataset:
+      name: SPGI Speech
+      type: kensho/spgispeech
+      config: test
+      split: test
+      args:
+        language: en
+    metrics:
+    - name: Test WER
+      type: wer
+      value: 1.9
+  - task:
+      type: Automatic Speech Recognition
+      name: automatic-speech-recognition
+    dataset:
+      name: tedlium-v3
+      type: LIUM/tedlium
+      config: release1
+      split: test
+      args:
+        language: en
+    metrics:
+    - name: Test WER
+      type: wer
+      value: 2.71
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: Vox Populi
+      type: facebook/voxpopuli
+      config: en
+      split: test
+      args:
+        language: en
+    metrics:
+    - name: Test WER
+      type: wer
+      value: 5.66
+metrics:
+- wer
+base_model:
+- nvidia/canary-1b-flash
+- Qwen/Qwen3-1.7B
+---
+
+<style>
+img {
+ display: inline;
+}
+</style>
+
+[![Model architecture](https://img.shields.io/badge/Model_Arch-SALM-blue#model-badge)](#model-architecture)
+| [![Model size](https://img.shields.io/badge/Params-2.5B-green#model-badge)](#model-architecture)
+| [![Language](https://img.shields.io/badge/Language-en-orange#model-badge)](#datasets)
+
+
+# Model Overview
+
+## Description:
+NVIDIA NeMo Canary-Qwen-2.5B is an English speech recognition model that achieves state-of-the art performance on multiple English speech benchmarks. With 2.5 billion parameters and running at 418 RTFx, Canary-Qwen-2.5B supports automatic speech-to-text recognition (ASR) in English with punctuation and capitalization (PnC). The model works in two modes: as a transcription tool (ASR mode) and as an LLM (LLM mode). In ASR mode, the model is only capable of transcribing the speech into text, but does not retain any LLM-specific skills such as reasoning. In LLM mode, the model retains all of the original LLM capabilities, which can be used to post-process the transcript, e.g. summarize it or answer questions about it. In LLM mode, the model does not "understand" the raw audio anymore - only its transcript. This model is ready for commercial use.
+
+### License/Terms of Use: 
+Canary-Qwen-2.5B is released under the CC-BY-4.0 license. By using this model, you are agreeing to the [terms and conditions](https://choosealicense.com/licenses/cc-by-4.0/) of the license. <br>
+
+### Discover more from NVIDIA: 
+For documentation, deployment guides, enterprise-ready APIs, and the latest open models—including Nemotron and other cutting-edge speech, translation, and generative AI—visit the NVIDIA Developer Portal at developer.nvidia.com.
+Join the community to access tools, support, and resources to accelerate your development with NVIDIA’s NeMo, Riva, NIM, and foundation models.<br>
+
+#### Explore more from NVIDIA: <br>
+What is [Nemotron](https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/)?<br>
+NVIDIA Developer [Nemotron](https://developer.nvidia.com/nemotron)<br>
+[NVIDIA Riva Speech](https://developer.nvidia.com/riva?sortBy=developer_learning_library%2Fsort%2Ffeatured_in.riva%3Adesc%2Ctitle%3Aasc#demos)<br>
+[NeMo Documentation](https://docs.nvidia.com/nemo-framework/user-guide/latest/nemotoolkit/asr/models.html)<br>
+
+## References:
+[1] [Less is More: Accurate Speech Recognition & Translation without Web-Scale Data](https://www.isca-archive.org/interspeech_2024/puvvada24_interspeech.pdf)
+
+[2] [Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition](https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10389701)
+
+[3] [Attention Is All You Need](https://arxiv.org/abs/1706.03762)
+
+[4] [Qwen/Qwen3-1.7B Model Card](https://huggingface.co/Qwen/Qwen3-1.7B)
+
+[5] [Training and Inference Efficiency of Encoder-Decoder Speech Models](https://arxiv.org/abs/2503.05931)
+
+[6] [NVIDIA NeMo Toolkit](https://github.com/NVIDIA/NeMo)
+
+[7] [Granary: Speech Recognition and Translation Dataset in 25 European Languages](https://arxiv.org/abs/2505.13404)
+
+[8] [Towards Measuring Fairness in AI: the Casual Conversations Dataset](https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9634168)
+
+[9] [SALM: Speech-augmented Language Model with In-context Learning for Speech Recognition and Translation](https://arxiv.org/abs/2310.09424) 
+
+### Deployment Geography:
+
+Global
+
+### Use Case:
+
+The model is intended for users requiring speech-to-text transcription capabilities for English speech, and/or transcript post-processing capabilities enabled by prompting the underlying LLMs. Typical use-cases: transcription, summarization, answering user questions about the transcript.
+
+### Release Date:
+
+Huggingface 07/17/2025 via https://huggingface.co/nvidia/canary-qwen-2.5b
+
+## Discover more from NVIDIA:
+For documentation, deployment guides, enterprise-ready APIs, and the latest open models—including Nemotron and other cutting-edge speech, translation, and generative AI—visit the NVIDIA Developer Portal at [developer.nvidia.com](https://developer.nvidia.com/).
+Join the community to access tools, support, and resources to accelerate your development with NVIDIA’s NeMo, Riva, NIM, and foundation models.<br>
+
+### Explore more from NVIDIA:  <br>
+What is [Nemotron](https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/)?<br>
+NVIDIA Developer [Nemotron](https://developer.nvidia.com/nemotron)<br>
+[NVIDIA Riva Speech](https://developer.nvidia.com/riva?sortBy=developer_learning_library%2Fsort%2Ffeatured_in.riva%3Adesc%2Ctitle%3Aasc#demos)<br>
+[NeMo Documentation](https://docs.nvidia.com/nemo-framework/user-guide/latest/nemotoolkit/asr/models.html)<br>
+
+## Model Architecture:
+Canary-Qwen is a Speech-Augmented Language Model (SALM) [9] model with FastConformer [2] Encoder and Transformer Decoder [3]. It is built using two base models: `nvidia/canary-1b-flash` [1,5] and `Qwen/Qwen3-1.7B` [4], a linear projection, and low-rank adaptation (LoRA) applied to the LLM. The audio encoder computes audio representation that is mapped to the LLM embedding space via a linear projection, and concatenated with the embeddings of text tokens. The model is prompted with "Transcribe the following: <audio>", using Qwen's chat template.
+
+### Limitations
+
+**Input length.** The maximum audio duration in training was 40s, and the maximum token sequence length was 1024 tokens (including prompt, audio, and response). The model may technically be able to process longer sequences, but its accuracy may be degraded.
+
+**Exclusively ASR oriented capabilities.** The model is not expected to preserve any of the underlying LLM's capabilities into speech modality. 
+
+**English-only language support.** The model was trained using English data only. It may be able to spuriously transcribe other languages as the underlying encoder was pretrained using German, French, and Spanish speech in addition to English, but it's unlikely to be reliable as a multilingual model.
+
+## NVIDIA NeMo
+
+To train, fine-tune or transcribe with Canary-Qwen-2.5B, you will need to install [NVIDIA NeMo](https://github.com/NVIDIA/NeMo).
+
+```bash
+# Currently requires installing the latest trunk version of NeMo, and PyTorch 2.6+ for FSDP2 support.
+python -m pip install "nemo_toolkit[asr,tts] @ git+https://github.com/NVIDIA/NeMo.git"
+```
+
+## How to Use this Model
+
+The model is available for use in the NVIDIA NeMo toolkit [6], and can be used as a pre-trained checkpoint for inference or for fine-tuning on another dataset.
+
+### Loading the Model
+
+```python
+from nemo.collections.speechlm2.models import SALM
+
+model = SALM.from_pretrained('nvidia/canary-qwen-2.5b')
+```
+
+## Input: 
+
+**Input Type(s):** Audio, text prompt <br>
+**Input Format(s):** Audio: .wav or .flac files. Text prompt string for ASR mode: `Transcribe the following: <|audioplaceholder|>` <br>
+**Input Parameters(s):** Audio: Two-Dimensional (batch, audio-samples); Text: One-Dimensional (string) <br>
+**Other Properties Related to Input:** 16000 Hz Mono-channel Audio, Pre-Processing Not Needed <br>
+
+Input to Canary-Qwen-2.5B is a batch of prompts that include audio.
+
+Example usage in ASR mode (speech-to-text):
+
+```python
+answer_ids = model.generate(
+    prompts=[
+        [{"role": "user", "content": f"Transcribe the following: {model.audio_locator_tag}", "audio": ["speech.wav"]}]
+    ],
+    max_new_tokens=128,
+)
+print(model.tokenizer.ids_to_text(answer_ids[0].cpu()))
+```
+
+Example usage in LLM mode (text-only):
+
+```python
+prompt = "..."
+transcript = "..."
+with model.llm.disable_adapter():
+    answer_ids = model.generate(
+        prompts=[[{"role": "user", "content": f"{prompt}\n\n{transcript}"}]],
+        max_new_tokens=2048,
+    )
+```
+
+To transcribe a dataset of recordings, specify the input as jsonl manifest file, where each line in the file is a dictionary containing the following fields: 
+
+```yaml
+# Example of a line in input_manifest.json
+{
+    "audio_filepath": "/path/to/audio.wav",  # path to the audio file
+    "duration": 30.0,  # duration of the audio
+}
+```
+
+and then use:
+```bash
+cd NeMo
+python examples/speechlm2/salm_generate.py \
+  pretrained_name=nvidia/canary-qwen-2.5b \
+  inputs=input_manifest.json \
+  output_manifest=generations.jsonl \
+  batch_size=128 \
+  user_prompt="Transcribe the following:"  # audio locator is added automatically at the end if not present
+```
+
+## Output:
+**Output Type(s):** Text <br>
+**Output Format:** Text transcript as a sequence of token IDs or a string <br> 
+**Output Parameters:** One-Dimensional text string <br>
+**Other Properties Related to Output:** May Need Inverse Text Normalization <br>
+
+Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA’s hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.
+
+## Software Integration:
+**Runtime Engine(s):** 
+* NeMo - 2.5.0 or higher <br>
+
+**Supported Hardware Microarchitecture Compatibility:** <br>
+* [NVIDIA Ampere] <br>
+* [NVIDIA Blackwell] <br>
+* [NVIDIA Jetson]  <br>
+* [NVIDIA Hopper] <br>
+* [NVIDIA Lovelace] <br>
+* [NVIDIA Pascal] <br>
+* [NVIDIA Turing] <br>
+* [NVIDIA Volta] <br>
+
+**[Preferred/Supported] Operating System(s):** <br>
+* [Linux] <br>
+* [Linux 4 Tegra] <br>
+* [Windows] <br>
+
+## Model Version(s): 
+Canary-Qwen-2.5B <br>
+
+## Training
+
+Canary-Qwen-2.5B was trained using the NVIDIA NeMo toolkit [6] for a total of 90k steps on 32 NVIDIA A100 80GB GPUs. LLM parameters were kept frozen. Speech encoder, projection, and LoRA parameters were trainable. The encoder's output frame rate is 80ms, or 12.5 tokens per second. The model was trained on approximately 1.3B tokens in total (this number inlcudes the speech encoder output frames, text response tokens, prompt tokens, and chat template tokens). 
+
+The model can be trained using this [example script](https://github.com/NVIDIA/NeMo/blob/main/examples/speechlm2/salm_train.py) and [base config](https://github.com/NVIDIA/NeMo/blob/main/examples/speechlm2/conf/salm.yaml).
+
+The tokenizer was inherited from `Qwen/Qwen3-1.7B`.
+
+# Training and Evaluation Datasets: 
+
+## Training Dataset:
+
+** The total size (in number of data points): approx. 40 million (speech, text) pairs
+** Total number of datasets: 26, with 18 for training and 8 for test
+** Dataset partition: Training 99.6%, testing 0.04%, validation 0%
+** Time period for training data collection: 1990-2025
+** Time period for testing data collection: 2005-2022
+** Time period for validation data collection N/A (unused)
+
+The Canary-Qwen-2.5B model is trained on a total of 234K hrs of publicly available speech data.
+The datasets below include conversations, videos from the web and audiobook recordings.
+
+**Data Collection Method:**
+* Human <br>
+
+**Labeling Method:**
+* Hybrid: Human, Automated <br>
+
+### Properties
+
+#### English (234.5k hours)
+
+The majority of the training data comes from the English portion of the Granary dataset [7]:
+
+- YouTube-Commons (YTC) (109.5k hours)
+- YODAS2 (77k hours)
+- LibriLight (13.6k hours)
+
+In addition, the following datasets were used:
+- Librispeech 960 hours
+- Fisher Corpus
+- Switchboard-1 Dataset
+- WSJ-0 and WSJ-1
+- National Speech Corpus (Part 1, Part 6)
+- VCTK
+- VoxPopuli (EN)
+- Europarl-ASR (EN)
+- Multilingual Librispeech (MLS EN)
+- Mozilla Common Voice (v11.0)
+- Mozilla Common Voice (v7.0)
+- Mozilla Common Voice (v4.0)
+- AMI
+- FLEURS
+
+AMI was oversampled during model training to constitute about 15% of the total data observed. 
+This skewed the model towards predicting verbatim transcripts that include conversational speech disfluencies such as repetitions.
+
+The training transcripts contained punctuation and capitalization.
+
+## Evaluation Dataset:
+
+**Data Collection Method:** <br>
+* Human <br>
+
+**Labeling Method:** <br>
+* Human <br>
+
+Automatic Speech Recognition: 
+* [HuggingFace OpenASR Leaderboard evaluation sets](https://huggingface.co/spaces/hf-audio/open_asr_leaderboard)
+
+Hallucination Robustness:
+* [MUSAN](https://www.openslr.org/17/) 48 hrs eval set
+
+Noise Robustness:
+* [Librispeech](https://www.openslr.org/12)
+
+Model Fairness:
+* [Casual Conversations Dataset](https://arxiv.org/pdf/2104.02821)
+
+## Performance
+
+The ASR predictions were generated using greedy decoding.
+
+### ASR Performance (w/o PnC) 
+
+The ASR performance is measured with word error rate (WER), and we process the groundtruth and predicted text with [whisper-normalizer](https://pypi.org/project/whisper-normalizer/) version 0.1.12.
+
+WER on [HuggingFace OpenASR leaderboard](https://huggingface.co/spaces/hf-audio/open_asr_leaderboard):
+
+| **Version** | **Model**     | **RTFx**   | **Mean**   | **AMI**   | **GigaSpeech**   | **LS Clean**   | **LS Other**   | **Earnings22**   | **SPGISpech**   | **Tedlium**   | **Voxpopuli**   |
+|:---------:|:-----------:|:------:|:------:|:------:|:------:|:------:|:------:|:------:|:------:|:------:|:------:|
+| 2.5.0  | Canary-Qwen-2.5B | 418 | 5.63 | 10.18 | 9.41 | 1.60 | 3.10 | 10.42 | 1.90 | 2.72 | 5.66 |
+
+More details on evaluation can be found at [HuggingFace ASR Leaderboard](https://huggingface.co/spaces/hf-audio/open_asr_leaderboard)
+
+### Hallucination Robustness
+Number of characters per minute on [MUSAN](https://www.openslr.org/17) 48 hrs eval set (`max_new_tokens=50` following `nvidia/canary-1b-flash` evaluation)
+| **Version** | **Model** | **# of character per minute** |
+|:-----------:|:---------:|:----------:|
+| 2.5.0       | Canary-Qwen-2.5B |   138.1   |
+
+### Noise Robustness
+WER on [Librispeech Test Clean](https://www.openslr.org/12) at different SNR (signal to noise ratio) levels of additive white noise
+
+| **Version** | **Model** | **SNR 10** | **SNR 5** | **SNR 0** | **SNR -5** |
+|:-----------:|:---------:|:----------:|:----------:|:----------:|:----------:|
+| 2.5.0       | Canary-Qwen-2.5B |    2.41%   |   4.08%   |   9.83%   |    30.60%  |
+
+## Model Fairness Evaluation
+
+As outlined in the paper "Towards Measuring Fairness in AI: the Casual Conversations Dataset" [8], we assessed the Canary-Qwen-2.5B model for fairness. The model was evaluated on the CasualConversations-v1 dataset with inference done on non-overlapping 40s chunks, and the results are reported as follows:
+
+### Gender Bias:
+
+| Gender | Male | Female | N/A | Other |
+| :--- | :--- | :--- | :--- | :--- |
+| Num utterances | 18471 | 23378 | 880 | 18 |
+| % WER | 16.71 | 13.85 | 17.71 | 29.46 |
+
+### Age Bias:
+
+| Age Group | (18-30) | (31-45) | (46-85) | (1-100) |
+| :--- | :--- | :--- | :--- | :--- |
+| Num utterances | 15058 | 13984 | 12810 | 41852 |
+| % WER | 15.73 | 15.3 | 14.14 | 15.11 |
+
+(Error rates for fairness evaluation are determined by normalizing both the reference and predicted text, similar to the methods used in the evaluations found at https://github.com/huggingface/open_asr_leaderboard.)
+
+## Inference:
+**Engine:** NVIDIA NeMo <br>
+**Test Hardware :** <br>
+* A6000 <br>
+* A100 <br>
+* RTX 5090 <br>
+
+## Ethical Considerations:
+NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications.  When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.  
+For more detailed information on ethical considerations for this model, please see the Model Card++ Explainability, Bias, Safety & Security, and Privacy Subcards. Please report security vulnerabilities or NVIDIA AI Concerns [here](https://www.nvidia.com/en-us/support/submit-security-vulnerability/).

config.json

@@ -0,0 +1,95 @@
+{
+  "audio_locator_tag": "<|audioplaceholder|>",
+  "freeze_params": [
+    "^llm\\..+$",
+    "^embed_tokens\\..+$"
+  ],
+  "lora": {
+    "lora_alpha": 256,
+    "lora_dropout": 0.01,
+    "r": 128,
+    "target_modules": [
+      "q_proj",
+      "v_proj"
+    ],
+    "task_type": "CAUSAL_LM"
+  },
+  "lr_scheduler": {
+    "_target_": "nemo.core.optim.lr_scheduler.CosineAnnealing",
+    "max_steps": 100000,
+    "min_lr": 1e-06,
+    "warmup_steps": 1000
+  },
+  "optimizer": {
+    "_target_": "torch.optim.AdamW",
+    "betas": [
+      0.9,
+      0.98
+    ],
+    "foreach": true,
+    "lr": 0.0005,
+    "weight_decay": 0.001
+  },
+  "perception": {
+    "encoder": {
+      "_target_": "nemo.collections.asr.modules.ConformerEncoder",
+      "att_context_size": [
+        -1,
+        -1
+      ],
+      "causal_downsampling": false,
+      "conv_context_size": null,
+      "conv_kernel_size": 9,
+      "conv_norm_type": "batch_norm",
+      "d_model": 1024,
+      "dropout": 0.1,
+      "dropout_att": 0.1,
+      "dropout_emb": 0.0,
+      "dropout_pre_encoder": 0.1,
+      "feat_in": 128,
+      "feat_out": -1,
+      "ff_expansion_factor": 4,
+      "n_heads": 8,
+      "n_layers": 32,
+      "pos_emb_max_len": 5000,
+      "reduction": null,
+      "reduction_factor": 1,
+      "reduction_position": null,
+      "self_attention_model": "rel_pos",
+      "subsampling": "dw_striding",
+      "subsampling_conv_channels": 256,
+      "subsampling_factor": 8,
+      "untie_biases": true,
+      "xscaling": false
+    },
+    "modality_adapter": {
+      "_target_": "nemo.collections.speechlm2.modules.perception.IdentityConnector",
+      "d_model": 1024
+    },
+    "output_dim": 2048,
+    "preprocessor": {
+      "_target_": "nemo.collections.asr.modules.AudioToMelSpectrogramPreprocessor",
+      "dither": 1e-05,
+      "features": 128,
+      "frame_splicing": 1,
+      "log": true,
+      "n_fft": 512,
+      "normalize": "per_feature",
+      "pad_to": 0,
+      "pad_value": 0.0,
+      "sample_rate": 16000,
+      "window": "hann",
+      "window_size": 0.025,
+      "window_stride": 0.01
+    },
+    "target": "nemo.collections.speechlm2.modules.perception.AudioPerceptionModule"
+  },
+  "pretrained_asr": "nvidia/canary-1b-flash",
+  "pretrained_llm": "Qwen/Qwen3-1.7B",
+  "pretrained_weights": false,
+  "prevent_freeze_params": [
+    "^.+\\.lora_.+$"
+  ],
+  "prompt_format": "qwen",
+  "torch_dtype": "bfloat16"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:800cb0d099cf655a8887d8b741c3a4afa9891e2b2949870251c4d58c72b59175
+size 5119120624