This model is a fine-tuned version of MIT/ast-finetuned-audioset-10-10-0.4593 on an unknown dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 1e-05 - trainbatchsize: 16 - evalbatchsize: 16 - lrschedulertype: linear - lrschedulerwarmupsteps: 10 - numepochs: 4 - mixedprecisiontraining: Native AMP - Transformers 5.16.1 - Pytorch 2.11.0+cu128 - Datasets 4.8.5 - Tokenizers 0.23.1
Open-weight model · Audio classification
ast-finetuned-audioset-10-10-0.4593
by Massachusetts Institute of Technology MIT/ast-finetuned-audioset-10-10-0.4593
ast-finetuned-audioset-10-10-0.4593 is an open-weight model for audio classification from Massachusetts Institute of Technology, released under BSD 3-Clause License. It has 87M parameters. At 16-bit it needs about 0.2 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 1.2M downloads a month.
Audio Spectrogram Transformer (AST) model fine-tuned on AudioSet. It was introduced in the paper AST: Audio Spectrogram Transformer by Gong et al. and first released in this repository.
Parameters87M
Context—
Weights692.9 MB
Licensebsd-3-clause
AccessOpen weights
Monthly Downloads1.2M
Runs On
What it takes to serve ast-finetuned-audioset-10-10-0.4593 (87M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.2 GB | 0.2 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.0 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 16, 2026.
ast-finetuned-audioset-10-10-0.4593 on every accelerator the SAVRN Index prices, at every precision
SAVRN's Notes on ast-finetuned-audioset-10-10-0.4593
87M parameters, 692.9 MB on disk, 0.2 GB at 16-bit, 687.5k downloads a month: the Audio Spectrogram Transformer fine-tuned on AudioSet is another model where the Index's cheapest rental, one MI300X at $1.85 an hour, is the wrong answer — this classifies audio tags on a CPU or any spare accelerator slice, and its 4-bit line rounds to 0.1 GB.
BSD 3-Clause permits commercial use with attribution intact — lighter notice burden than Apache, no patent grant language to track. The weights ship as two 346.4 MB files (safetensors and pytorch; take one), and the lineage is documented in arXiv:2104.01778. Fine-tunes of this very checkpoint, like the beewatch_ast_3class neighbor on this hub, are common, which tells you it works well as a base for custom audio labels. No evaluations are reported on this page, so score it on your own clips.
Model Card
By Massachusetts Institute of Technology, published under bsd-3-clause, revision f826b80d2822.
Audio Spectrogram Transformer (AST) model fine-tuned on AudioSet. It was introduced in the paper AST: Audio Spectrogram Transformer by Gong et al. and first released in this repository. Disclaimer: The team releasing Audio Spectrogram Transformer did not write a model card for this model so this model card has been written by the Hugging Face team. The Audio Spectrogram Transformer is equivalent to ViT, but applied on audio. Audio is first turned into an image (as a spectrogram), after which a Vision Transformer is applied. The model gets state-of-the-art results on several audio classification benchmarks. You can use the raw model for classifying audio into one of the AudioSet classes. See…
Read Massachusetts Institute of Technology's full model card
Audio Spectrogram Transformer (fine-tuned on AudioSet)
Audio Spectrogram Transformer (AST) model fine-tuned on AudioSet. It was introduced in the paper AST: Audio Spectrogram Transformer by Gong et al. and first released in this repository.
Disclaimer: The team releasing Audio Spectrogram Transformer did not write a model card for this model so this model card has been written by the Hugging Face team.
Model description
The Audio Spectrogram Transformer is equivalent to ViT, but applied on audio. Audio is first turned into an image (as a spectrogram), after which a Vision Transformer is applied. The model gets state-of-the-art results on several audio classification benchmarks.
Usage
You can use the raw model for classifying audio into one of the AudioSet classes. See the documentation for more info.
Configuration
- Architecture
- ASTForAudioClassification
- Layers
- 12
- Hidden size
- 768
- Feed-forward size
- 3,072
- Attention heads
- 12
- Stored precision
- float32
- Model type
- audio-spectrogram-transformer
Identity and Version
- Repository
- MIT/ast-finetuned-audioset-10-10-0.4593
- Publisher
- Massachusetts Institute of Technology
- Task
- Audio classification
- Modality
- Audio
- Library
- transformers
- Parameters
- 87M parameters
- Languages
- Not stated by the source
- Revision
- f826b80d28226b62986cc218e5cec390b1096902
- First published
- 2022-11-14
- Last updated
- 2023-09-06
Files and Weights
6 files, 692.9 MB in total. The weights are 2 files totalling 692.9 MB in bin, safetensors.
Weights2 files · 692.9 MB
Configuration2 files · 27.1 KB
Documentation1 file · 1.2 KB
Repository1 file · 1.5 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 346.4 MB | ae0c1e2ad4e1 |
| pytorch_model.bin | Weights | 346.4 MB | 9ca280ba0276 |
| config.json | Configuration | 26.8 KB | — |
| preprocessor_config.json | Configuration | 297 B | — |
| README.md | Documentation | 1.2 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- bsd-3-clause
- Access
- Open weights, no gate
- Download size
- 692.9 MB
Download from Massachusetts Institute of Technology
Released by Massachusetts Institute of Technology through its official repository on Hugging Face. Read the license.
Built From
- Described by arXiv:2104.01778
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 692.9 MB |
| 16-bit | 0.2 GB |
| 8-bit | 0.1 GB |
| 4-bit | 0.0 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Built on This Model
- Quantized fromast-finetuned-audioset-10-10-0.4593
- Derived fromast-finetuned-audioset-10-10-0.4593
- Derived frombeewatch_ast_3class
Questions About ast-finetuned-audioset-10-10-0.4593
How much GPU memory does ast-finetuned-audioset-10-10-0.4593 need?
About 0.2 GB at 16-bit and 0.1 GB at 4-bit: the weights (87M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run ast-finetuned-audioset-10-10-0.4593 on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use ast-finetuned-audioset-10-10-0.4593 commercially?
Yes. ast-finetuned-audioset-10-10-0.4593 is released under BSD 3-Clause License. The BSD 3-Clause License is permissive. It permits commercial use and redistribution with the copyright notice, and forbids using the authors' names to endorse derived products without permission.
Similar Models
Model · Audio classification
ast-finetuned-audioset-14-14-0.443
Audio Spectrogram Transformer (AST) model fine-tuned on AudioSet. It was introduced in the paper AST: Audio Spectrogram Transformer by Gong et al. and first released in this repository. Disclaimer: The team releasing Audio Spectrogram Transformer did not write a model card for this model so this model card has been written by the Hugging Face team. The Audio Spectrogram Transformer is equivalent to ViT, but applied on audio. Audio is first turned into an image (as a spectrogram), after which a Vision Transformer is applied. The model gets state-of-the-art results on several audio classification benchmarks. You can use the raw model for classifying audio into one of the AudioSet classes. See…
Model · Audio classification
discogs-maest-10s-pw-129e
MAEST is a family of Transformer models based on PASST and focused on music analysis applications. The MAEST models are also available for inference in the Essentia library and for inference and training in the official repository. You can try the MAEST interactive demo on replicate. MAEST is a music audio representation model pre-trained on the task of music style classification. According to the evaluation reported in the original paper, it reports good performance in several downstream music analysis tasks. The MAEST models can make predictions for a taxonomy of 400 music styles derived from the public metadata of Discogs. The MAEST models have reported good performance in downstream…
CED are simple ViT-Transformer-based models for audio tagging, achieving sota performance on Audioset. Notable differences from other available models include: 1. Simplification for finetuning: Batchnormalization of Mel-Spectrograms. During finetuning one does not need to first compute mean/variance over the dataset, which is common for AST. 1. Support for variable length inputs. Most other models use a static time-frequency position embedding, which hinders the model's generalization to segments shorter than 10s. Many previous transformers simply pad their input to 10s in order to avoid the performance impact, which in turn slows down training/inference drastically. 1. Training/Inference…
Dasheng (Deep Audio-Signal Holistic Embeddings), or “大声” ("great sound"), is a general-purpose audio encoder trained on a large-scale self-supervised learning task. Dasheng is designed to capture rich audio information across various domains, including speech, music, and environmental sounds. The model is trained on 272,356 hours of diverse audio data with 1.2 billion parameters, and exhibits significant performance gains on the HEAR benchmark. Dasheng outperforms previous works on CREMA-D, LibriCount, Speech Commands, VoxLingua, and competes well in music and environmental sound classification tasks. examplefinetuneesc50.ipynb demonstrates how to train a linear head on the ESC-50 dataset…
Audio Spectrogram Transformer (AST) model fine-tuned on Speech Commands v2. It was introduced in the paper AST: Audio Spectrogram Transformer by Gong et al. and first released in this repository. Disclaimer: The team releasing Audio Spectrogram Transformer did not write a model card for this model so this model card has been written by the Hugging Face team. The Audio Spectrogram Transformer is equivalent to ViT, but applied on audio. Audio is first turned into an image (as a spectrogram), after which a Vision Transformer is applied. The model gets state-of-the-art results on several audio classification benchmarks. You can use the raw model for classifying audio into one of the Speech…