# ONNX Format

Qwen3-1.7B is a 1.7B-parameter open-source large language model released by Alibaba Cloud, based on the Transformer architecture, supporting various natural language processing tasks.

NVIDIA FastConformer-Hybrid Large is a Russian automatic speech recognition model based on the FastConformer architecture, supporting CTC and RNN-T decoders.

A lightweight zero-shot object detection model in ONNX format, compatible with Transformers.js, suitable for browser-side deployment.

IBM Granite series time series forecasting model, based on PatchTST architecture, suitable for various time series forecasting tasks.

A lightweight portrait segmentation model in ONNX format, specifically optimized for separating people from backgrounds in landscape images.

TimeSformer-HR is a video action recognition model optimized for high-resolution videos and fine-tuned on the Kinetics-600 dataset.

TimeSformer is a Transformer-based video understanding model specifically optimized for temporal action recognition tasks.

TimeSformer is a video understanding model based on the Transformer architecture, specifically designed for video classification tasks.

TimeSformer is a Transformer-based video understanding model, specifically fine-tuned on the Kinetics-400 dataset.

Whisper is a general-purpose speech recognition model trained by OpenAI. This model is based on large-scale weakly supervised training and supports speech transcription in multiple languages.

Whisper is an automatic speech recognition (ASR) system trained by OpenAI, supporting multilingual speech transcription.

GAN-based 4x super-resolution image upscaling model, compatible with Transformers.js

This is the ONNX format conversion version of the gyr66/bert-base-chinese-finetuned-ner model, designed for Chinese named entity recognition tasks.

ONNX version of depth estimation model based on Transformers.js, suitable for web applications

This is the ONNX format conversion of the fmops/distilbert-prompt-injection model, designed for detecting prompt injection attacks.

This is the ONNX format version of the dslim/bert-base-NER model for named entity recognition tasks, capable of identifying four entity types: location, organization, person, and miscellaneous.

Hybrid depth estimation model developed by Intel, combining the advantages of convolutional neural networks and Transformer architecture

Lightweight Swin2SR image super-resolution model that can upscale image resolution by 2 times

A classical image super-resolution model based on Swin2SR architecture, capable of upscaling image resolution by 2 times

A Transformer-based handwritten text recognition model that converts handwritten images into text

A small Transformer-based handwritten text recognition model optimized for web usage

TrOCR is a Transformer-based OCR model specifically designed for recognizing printed text.

TrOCR-small-printed is a compact optical character recognition (OCR) model specifically designed for printed text recognition.

A German text embedding model based on sentence-transformers, mapping text to a 1024-dimensional vector space, specifically designed to enhance few-shot text classification performance in German

DistilBART-CNN-12-6 is a distilled version of the BART model, optimized for text summarization tasks, with a smaller size while maintaining high performance.

YOLOS-small is a small object detection model based on the Transformer architecture, designed for efficient visual tasks.

YOLOS-tiny is a lightweight object detection model based on the Transformer architecture, suitable for real-time object detection tasks.

TinyStories-1M-ONNX is a small language model based on the ONNX format, suitable for text generation tasks.

E5-small-v2 is an efficient text embedding model suitable for various natural language processing tasks.

ONNX format conversion of Facebook's wav2vec2-base-960h model, designed for Transformers.js, supporting browser-side speech recognition

MMS-LID-4017 is a speech recognition model supporting 4017 languages, developed by Facebook, focusing on language identification tasks.

MMS-LID-126 is a multilingual speech recognition model released by Facebook, supporting recognition of 126 languages.

A voice command recognition model based on AST architecture, optimized for web deployment in ONNX format

Audio Spectrogram Transformer (AST) model fine-tuned on the AudioSet dataset for audio classification tasks

Whisper Medium is a medium-scale speech recognition model developed by OpenAI, supporting automatic speech recognition (ASR) tasks in multiple languages.

End-to-end object detection model based on Transformer architecture with ResNet-101 feature extractor

Whisper Small is a small automatic speech recognition (ASR) model developed by OpenAI, capable of converting speech into text.

Whisper is an automatic speech recognition (ASR) system trained by OpenAI, supporting speech-to-text tasks in multiple languages.

Whisper Tiny is a lightweight speech recognition model open-sourced by OpenAI, suitable for web deployment.

A large-scale text summarization model based on the BART architecture, optimized for the CNN/DailyMail dataset