Skip to content

EthenEthenEthen

LoginPlatformTry Ethen

Open Source Model Profile · amaye15

Beit-Base-Image-Orientation-Fixer

Beit-Base-Image-Orientation-Fixer is an 86M-parameter BEiT image-classification checkpoint from amaye15. According to the model card, it is a fine-tuned orientation-fixer variant reporting evaluation loss 0.1295 and F1 0.9391.

Publisher
amaye15
Task
image-classification
Model type
beit
License
apache-2.0
Library
transformers
Publication status
Accepted · not indexed

Model overview

Beit-Base-Image-Orientation-Fixer is published by amaye15 as an image-classification model. Captured configuration identifies BeitForImageClassification with model type beit, and Safetensors metadata reports 85,978,180 parameters. According to the model card, it is a fine-tuned version of microsoft/swinv2-base-patch4-window16-256, while hub tags separately record a base-model reference to the same Swin v2 checkpoint.

Recorded capabilities

Orientation-fixer fine-tune claim

According to the model card, the checkpoint is a fine-tuned version of microsoft/swinv2-base-patch4-window16-256 for image-orientation fixing.

Reported F1 0.9391

According to the model card, it achieves loss 0.1295 and F1 0.9391 on the evaluation set; no independent evaluation scores were extracted.

86M-parameter BEiT config

Captured configuration identifies BeitForImageClassification with model type beit, and Safetensors metadata reports 85,978,180 parameters.

Documented training hyperparameters

According to the model card, training used learning rate 5e-05, train and eval batch size 32, Adam with linear scheduling, seed 42, and 10,000 training steps.

Use cases in the source record

  • Image-orientation classification for fixing image rotation, matching the publisher's orientation-fixer description and image-classification pipeline.
  • Transformers-based image-classification experiments using the Safetensors checkpoint with the documented training setup of learning rate 5e-05 and batch size 32.

Limitations and unknowns

  • According to the model card, intended uses, limitations, and training and evaluation data are recorded as needing more information, and the dataset is listed as None.
  • No independent evaluation results were extracted; the only reported loss and F1 figures are publisher claims.
  • No quantization, context-window, or hardware requirement was extracted.
  • Provider state is historical snapshot data, not independently refreshed current availability.

Source and provenance

Source: amaye15/Beit-Base-Image-Orientation-Fixer

Captured: Unknown. Processed: 2026-09-07T19:34:39.809169+00:00.

SwinV2-Base-Image-Orientation-Fixer This model is a fine-tuned version of microsoft/swinv2-base-patch4-window16-256 on the None dataset. It achieves the following results on the evaluation set: Loss: 0.1295 F1: 0.9391 Model description More information needed Intended uses & limitations More information needed Training and evaluation data More information needed Training procedure Training hyperparameters The following hyperparameters were used during training: learning_rate: 5e-05 train_batch_size: 32 eval_batch_size: 32 seed: 42 optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 lr_scheduler_type: linear training_steps: 1000…

F001F002F003F004F005F006F007F008F009F010F011F012F013