From 64cb68384241f046d07ba3797e1b14dbc82a0b2d Mon Sep 17 00:00:00 2001 From: Sohaib Ahmed <113179991+Sohaib-Ahmed21@users.noreply.github.com> Date: Mon, 5 Oct 2026 21:28:09 +0500 Subject: [PATCH] [Model] Add Nemotron 3.5 ASR transcription support (#59827) Signed-off-by: Sohaib-Ahmed21 --- docs/models/supported_models.md | 1 + docs/usage/v1_guide.md | 1 + vllm/model_executor/models/registry.py | 1 + 3 files changed, 3 insertions(+) diff --git a/docs/models/supported_models.md b/docs/models/supported_models.md index a55559b0d6..221ab877ea 100644 --- a/docs/models/supported_models.md +++ b/docs/models/supported_models.md @@ -199,6 +199,7 @@ Some model architectures are supported via vLLM plugins. These plugins extend vL | ------------ | ------ | ----------------- | | `BartForConditionalGeneration` | BART | [bart-plugin](https://github.com/vllm-project/bart-plugin) | | `Florence2ForConditionalGeneration` | Florence-2 | [bart-plugin](https://github.com/vllm-project/bart-plugin) | +| `Nemotron3_5AsrForRNNT` | NVIDIA Nemotron 3.5 ASR | [vllm-nemotron-asr-plugin](https://github.com/Sohaib-Ahmed21/vllm-nemotron-asr-plugin) | For other model architectures not natively supported, in particular for Encoder-Decoder models, we recommend following a similar pattern by implementing support through the plugin system. diff --git a/docs/usage/v1_guide.md b/docs/usage/v1_guide.md index 4ef2b0b5b6..d16db46895 100644 --- a/docs/usage/v1_guide.md +++ b/docs/usage/v1_guide.md @@ -138,6 +138,7 @@ Whisper is supported natively. Other encoder-decoder models are supported via th - **BART**: `BartForConditionalGeneration` is supported via the official [bart-plugin](https://github.com/vllm-project/bart-plugin). - **Florence-2**: `Florence2ForConditionalGeneration` is supported via the official [bart-plugin](https://github.com/vllm-project/bart-plugin). +- **Nemotron 3.5 ASR**: `Nemotron3_5AsrForRNNT` is supported via [vllm-nemotron-asr-plugin](https://github.com/Sohaib-Ahmed21/vllm-nemotron-asr-plugin). For other encoder-decoder models (e.g., `MllamaForConditionalGeneration`), we recommend following a similar pattern by implementing support through the [plugin system](../design/plugin_system.md). diff --git a/vllm/model_executor/models/registry.py b/vllm/model_executor/models/registry.py index 3ff37a2562..48b831bbdf 100644 --- a/vllm/model_executor/models/registry.py +++ b/vllm/model_executor/models/registry.py @@ -849,6 +849,7 @@ _OOT_SUPPORTED_MODELS = { "BartForConditionalGeneration": "https://github.com/vllm-project/bart-plugin", "Florence2ForConditionalGeneration": "https://github.com/vllm-project/bart-plugin", "MBartForConditionalGeneration": "https://github.com/vllm-project/bart-plugin", + "Nemotron3_5AsrForRNNT": "https://github.com/Sohaib-Ahmed21/vllm-nemotron-asr-plugin", }