Multi-label image classification datasets compatible with timm (>= v1.0.31)
AI & ML interests
Computer Vision
Recent Activity
-
timm/vit_base_patch16_sapiens2.fb
Image Feature Extraction • 0.1B • Updated • 345 -
timm/vit_large_patch16_sapiens2.fb
Image Feature Extraction • 0.4B • Updated • 306 -
timm/vit_huge_patch16_sapiens2.fb
Image Feature Extraction • 0.8B • Updated • 235 -
timm/vit_giant_patch16_sapiens2.fb
Image Feature Extraction • 1B • Updated • 318
Meta AI's DINOv3 weights in timm. ViTs with `qkvb` have a zero QV bias present, otherwise bias is disabled. QKV bias are all 0 in original weights.
-
timm/vit_7b_patch16_dinov3.sat493m
Image Feature Extraction • 7B • Updated • 961 • 1 -
timm/vit_7b_patch16_dinov3.lvd1689m
Image Feature Extraction • 7B • Updated • 22.6k -
timm/vit_huge_plus_patch16_dinov3.lvd1689m
Image Feature Extraction • 0.8B • Updated • 22.1k • 6 -
timm/vit_huge_plus_patch16_dinov3_qkvb.lvd1689m
Image Feature Extraction • 0.8B • Updated • 868
OpenCLIP / timm ports of Apple's MobileCLIP-2 multi-modal and image encoders
-
timm/MobileCLIP2-S4-OpenCLIP
Zero-Shot Image Classification • Updated • 2.39k • 3 -
timm/MobileCLIP2-S3-OpenCLIP
Zero-Shot Image Classification • Updated • 6.37k • 3 -
timm/MobileCLIP2-S2-OpenCLIP
Zero-Shot Image Classification • Updated • 73.5k • 6 -
timm/MobileCLIP2-S0-OpenCLIP
Zero-Shot Image Classification • Updated • 28k • 2
Exploring ViT hparams and model shapes for the GPU poor (between tiny and base).
-
timm/vit_so150m2_patch16_reg1_gap_448.sbb_e200_in12k_ft_in1k
Image Classification • 0.1B • Updated • 107 • 1 -
timm/vit_so150m2_patch16_reg1_gap_384.sbb_e200_in12k_ft_in1k
Image Classification • 0.1B • Updated • 48 • 2 -
timm/vit_so150m2_patch16_reg1_gap_256.sbb_e200_in12k_ft_in1k
Image Classification • 0.1B • Updated • 76 • 1 -
timm/vit_so150m2_patch16_reg1_gap_256.sbb_e200_in12k
Image Classification • 0.1B • Updated • 56 • 1
Weights for MobileNet-V4 pretrained in timm
-
timm/mobilenetv4_conv_aa_large.e230_r448_in12k_ft_in1k
Image Classification • 32.7M • Updated • 1.67k • 4 -
timm/mobilenetv4_conv_aa_large.e230_r384_in12k_ft_in1k
Image Classification • 32.7M • Updated • 220 • 1 -
timm/mobilenetv4_hybrid_large.ix_e600_r384_in1k
Image Classification • 37.9M • Updated • 270 • 5 -
timm/mobilenetv4_hybrid_large.e600_r384_in1k
Image Classification • 37.9M • Updated • 1.43k • 2
Not the most accurate, but the highest throughput image classification models in timm
-
timm/tinynet_e.in1k
Image Classification • 2.06M • Updated • 3.59k -
timm/mobilenetv3_small_050.lamb_in1k
Image Classification • 1.6M • Updated • 24.3k -
timm/lcnet_050.ra2_in1k
Image Classification • 1.89M • Updated • 6.01k -
timm/mobilenetv3_small_075.lamb_in1k
Image Classification • 2.05M • Updated • 12.1k • 1
timm includes the most popular convolutional and vision transformer models, many with new weights from updated training recipes.
Fastest image classification models with 80% accuracy in ImageNet-1k .
-
timm/levit_256.fb_dist_in1k
Image Classification • 19M • Updated • 58k • 2 -
timm/vit_base_patch32_clip_224.laion2b_ft_in1k
Image Classification • 88.2M • Updated • 138 • 1 -
timm/vit_base_patch32_clip_224.laion2b_ft_in12k_in1k
Image Classification • 88.2M • Updated • 1.04k • 4 -
timm/vit_base_patch32_clip_224.openai_ft_in1k
Image Classification • 88.2M • Updated • 503
Fastest image classification models with 86% accuracy in ImageNet-1k .
-
timm/vit_base_patch16_clip_224.laion2b_ft_in12k_in1k
Image Classification • 86.6M • Updated • 4.3k • 2 -
timm/beitv2_base_patch16_224.in1k_ft_in22k_in1k
Image Classification • 87M • Updated • 4.03k -
timm/convnext_base.clip_laion2b_augreg_ft_in12k_in1k
Image Classification • 88.6M • Updated • 107k -
timm/convnext_base.clip_laion2b_augreg_ft_in1k
Image Classification • 88.6M • Updated • 1.27k
Pre-trained feature extraction backbones available in timm.
-
timm/vit_small_patch14_dinov2.lvd142m
Image Feature Extraction • 22.1M • Updated • 1.37M • 8 -
timm/vit_large_patch14_dinov2.lvd142m
Image Feature Extraction • 0.3B • Updated • 210k • 17 -
timm/vit_base_patch16_224.dino
Image Feature Extraction • 85.8M • Updated • 75.3k • 6 -
timm/vit_base_patch16_clip_224.openai
Image Feature Extraction • Updated • 156k • 12
Datasets for fine-tune benchmarking, hparam tuning. All vetted and tested with timm scripts.
-
timm/deepseek_vit_412m.deepseek_v4_1_flash
Image Feature Extraction • 0.4B • Updated • 300 • 2 -
timm/deepseek_vit_412m.deepseek_v4_flash_vision_exp
Image Feature Extraction • 0.4B • Updated • 25 -
timm/deepseek_vit_412m_align.deepseek_v4_1_flash
Image Feature Extraction • 0.5B • Updated • 51 -
timm/deepseek_vit_412m_enc.deepseek_v4_1_flash
Image Feature Extraction • 0.5B • Updated • 101
-
timm/qwen3_vit_88m.qwen3_5_0_8b
Image Feature Extraction • 87.4M • Updated • 797 -
timm/qwen3_vit_88m_enc.qwen3_5_0_8b
Image Feature Extraction • 0.1B • Updated • 61 -
timm/qwen3_vit_88m_merge.qwen3_5_0_8b
Image Feature Extraction • 0.1B • Updated • 361 • 2 -
timm/qwen3_vit_306m.qwen3_vl_4b
Image Feature Extraction • 0.3B • Updated • 600
OpenCLIP (PE Core image + text) and timm PE Core, Spatial, Lang (ViT only) weights. NOTE: These weights do not work with original modeling code.
OpenCLIP and timm SigLIP 2 models
-
timm/ViT-SO400M-16-SigLIP2-naflex
Zero-Shot Image Classification • Updated • 73 • 1 -
timm/ViT-B-16-SigLIP2-naflex
Zero-Shot Image Classification • Updated • 87 -
timm/ViT-gopt-16-SigLIP2-384
Zero-Shot Image Classification • Updated • 4.25k • 4 -
timm/ViT-gopt-16-SigLIP2-256
Zero-Shot Image Classification • Updated • 2.01k
MetaCLIP & MetaCLIP2 OpenCLIP and timm models. All models are dual timm + OpenCLIP (or just timm for specific vit encoders).
-
timm/vit_gigantic_patch14_clip_378.metaclip2_worldwide
Zero-Shot Image Classification • 2B • Updated • 168 • 2 -
timm/vit_gigantic_patch14_clip_224.metaclip2_worldwide
Zero-Shot Image Classification • 2B • Updated • 90 • 1 -
timm/vit_huge_patch14_clip_378.metaclip2_worldwide
Zero-Shot Image Classification • 0.6B • Updated • 241 • 1 -
timm/vit_huge_patch14_clip_224.metaclip2_worldwide
Zero-Shot Image Classification • 0.6B • Updated • 606 • 1
The 20 best models on ImageNet-1k validation set, all pretrained on datasets larger than ImageNet and fine-tuned on ImageNet-1k.
-
timm/eva02_large_patch14_448.mim_m38m_ft_in22k_in1k
Image Classification • 0.3B • Updated • 16.7k • 25 -
timm/eva02_large_patch14_448.mim_in22k_ft_in22k_in1k
Image Classification • 0.3B • Updated • 2.76k • 1 -
timm/eva_giant_patch14_560.m30m_ft_in22k_in1k
Image Classification • 1B • Updated • 2.69k • 3 -
timm/eva02_large_patch14_448.mim_m38m_ft_in1k
Image Classification • 0.3B • Updated • 4.52k • 14
timm has a number of unique and exclusive models trained on a 11821 (12k) subset of the full ImageNet-22k
-
timm/convnext_xxlarge.clip_laion2b_soup_ft_in12k
Image Classification • 0.9B • Updated • 133 • 2 -
timm/vit_huge_patch14_clip_224.laion2b_ft_in12k
Image Classification • 0.6B • Updated • 62 • 1 -
timm/vit_large_patch14_clip_224.openai_ft_in12k
Image Classification • 0.3B • Updated • 40 -
timm/vit_large_patch14_clip_224.laion2b_ft_in12k
Image Classification • 0.3B • Updated • 66
Fastest image classification models with 75.3% accuracy in ImageNet-1k .
-
timm/levit_128s.fb_dist_in1k
Image Classification • 7.82M • Updated • 564 • 2 -
timm/vit_small_patch32_224.augreg_in21k_ft_in1k
Image Classification • 22.9M • Updated • 802 • 2 -
timm/levit_128.fb_dist_in1k
Image Classification • 9.26M • Updated • 6.22k • 1 -
timm/efficientvit_m5.r224_in1k
Image Classification • 12.5M • Updated • 439
Fastest image classification models with 83% accuracy in ImageNet-1k .
-
timm/vit_base_patch32_clip_224.laion2b_ft_in12k_in1k
Image Classification • 88.2M • Updated • 1.04k • 4 -
timm/deit3_small_patch16_224.fb_in22k_ft_in1k
Image Classification • 22.1M • Updated • 4.41k -
timm/tiny_vit_11m_224.dist_in22k_ft_in1k
Image Classification • 11M • Updated • 7.2k -
timm/tresnet_m.miil_in21k_ft_in1k
Image Classification • 31.5M • Updated • 255
Fastest image classification models with 88% accuracy in ImageNet-1k .
-
timm/eva_large_patch14_196.in22k_ft_in22k_in1k
Image Classification • 0.3B • Updated • 12.1k • 3 -
timm/beitv2_large_patch16_224.in1k_ft_in22k_in1k
Image Classification • 0.3B • Updated • 3.03k • 3 -
timm/vit_large_patch14_clip_224.openai_ft_in12k_in1k
Image Classification • 0.3B • Updated • 1.89k • 38 -
timm/convnext_large_mlp.clip_laion2b_soup_ft_in12k_in1k_384
Image Classification • 0.2B • Updated • 5.47k • 4
Noteworthy instances of ImageNet on the Hub. Vetted and tested with timm train and validation scripts.
A collection of very small (~300-500k parameter) models at 160x160 resolution, for testing purposes. Trained on ImageNet-1k.
-
timm/test_byobnet.r160_in1k
Image Classification • 459k • Updated • 1.04k • 1 -
timm/test_convnext.r160_in1k
Image Classification • 272k • Updated • 833 • 1 -
timm/test_convnext2.r160_in1k
Image Classification • 478k • Updated • 877 -
timm/test_convnext3.r160_in1k
Image Classification • 469k • Updated • 831
Multi-label image classification datasets compatible with timm (>= v1.0.31)
-
timm/deepseek_vit_412m.deepseek_v4_1_flash
Image Feature Extraction • 0.4B • Updated • 300 • 2 -
timm/deepseek_vit_412m.deepseek_v4_flash_vision_exp
Image Feature Extraction • 0.4B • Updated • 25 -
timm/deepseek_vit_412m_align.deepseek_v4_1_flash
Image Feature Extraction • 0.5B • Updated • 51 -
timm/deepseek_vit_412m_enc.deepseek_v4_1_flash
Image Feature Extraction • 0.5B • Updated • 101
-
timm/vit_base_patch16_sapiens2.fb
Image Feature Extraction • 0.1B • Updated • 345 -
timm/vit_large_patch16_sapiens2.fb
Image Feature Extraction • 0.4B • Updated • 306 -
timm/vit_huge_patch16_sapiens2.fb
Image Feature Extraction • 0.8B • Updated • 235 -
timm/vit_giant_patch16_sapiens2.fb
Image Feature Extraction • 1B • Updated • 318
-
timm/qwen3_vit_88m.qwen3_5_0_8b
Image Feature Extraction • 87.4M • Updated • 797 -
timm/qwen3_vit_88m_enc.qwen3_5_0_8b
Image Feature Extraction • 0.1B • Updated • 61 -
timm/qwen3_vit_88m_merge.qwen3_5_0_8b
Image Feature Extraction • 0.1B • Updated • 361 • 2 -
timm/qwen3_vit_306m.qwen3_vl_4b
Image Feature Extraction • 0.3B • Updated • 600
Meta AI's DINOv3 weights in timm. ViTs with `qkvb` have a zero QV bias present, otherwise bias is disabled. QKV bias are all 0 in original weights.
-
timm/vit_7b_patch16_dinov3.sat493m
Image Feature Extraction • 7B • Updated • 961 • 1 -
timm/vit_7b_patch16_dinov3.lvd1689m
Image Feature Extraction • 7B • Updated • 22.6k -
timm/vit_huge_plus_patch16_dinov3.lvd1689m
Image Feature Extraction • 0.8B • Updated • 22.1k • 6 -
timm/vit_huge_plus_patch16_dinov3_qkvb.lvd1689m
Image Feature Extraction • 0.8B • Updated • 868
OpenCLIP (PE Core image + text) and timm PE Core, Spatial, Lang (ViT only) weights. NOTE: These weights do not work with original modeling code.
OpenCLIP / timm ports of Apple's MobileCLIP-2 multi-modal and image encoders
-
timm/MobileCLIP2-S4-OpenCLIP
Zero-Shot Image Classification • Updated • 2.39k • 3 -
timm/MobileCLIP2-S3-OpenCLIP
Zero-Shot Image Classification • Updated • 6.37k • 3 -
timm/MobileCLIP2-S2-OpenCLIP
Zero-Shot Image Classification • Updated • 73.5k • 6 -
timm/MobileCLIP2-S0-OpenCLIP
Zero-Shot Image Classification • Updated • 28k • 2
OpenCLIP and timm SigLIP 2 models
-
timm/ViT-SO400M-16-SigLIP2-naflex
Zero-Shot Image Classification • Updated • 73 • 1 -
timm/ViT-B-16-SigLIP2-naflex
Zero-Shot Image Classification • Updated • 87 -
timm/ViT-gopt-16-SigLIP2-384
Zero-Shot Image Classification • Updated • 4.25k • 4 -
timm/ViT-gopt-16-SigLIP2-256
Zero-Shot Image Classification • Updated • 2.01k
Exploring ViT hparams and model shapes for the GPU poor (between tiny and base).
-
timm/vit_so150m2_patch16_reg1_gap_448.sbb_e200_in12k_ft_in1k
Image Classification • 0.1B • Updated • 107 • 1 -
timm/vit_so150m2_patch16_reg1_gap_384.sbb_e200_in12k_ft_in1k
Image Classification • 0.1B • Updated • 48 • 2 -
timm/vit_so150m2_patch16_reg1_gap_256.sbb_e200_in12k_ft_in1k
Image Classification • 0.1B • Updated • 76 • 1 -
timm/vit_so150m2_patch16_reg1_gap_256.sbb_e200_in12k
Image Classification • 0.1B • Updated • 56 • 1
MetaCLIP & MetaCLIP2 OpenCLIP and timm models. All models are dual timm + OpenCLIP (or just timm for specific vit encoders).
-
timm/vit_gigantic_patch14_clip_378.metaclip2_worldwide
Zero-Shot Image Classification • 2B • Updated • 168 • 2 -
timm/vit_gigantic_patch14_clip_224.metaclip2_worldwide
Zero-Shot Image Classification • 2B • Updated • 90 • 1 -
timm/vit_huge_patch14_clip_378.metaclip2_worldwide
Zero-Shot Image Classification • 0.6B • Updated • 241 • 1 -
timm/vit_huge_patch14_clip_224.metaclip2_worldwide
Zero-Shot Image Classification • 0.6B • Updated • 606 • 1
Weights for MobileNet-V4 pretrained in timm
-
timm/mobilenetv4_conv_aa_large.e230_r448_in12k_ft_in1k
Image Classification • 32.7M • Updated • 1.67k • 4 -
timm/mobilenetv4_conv_aa_large.e230_r384_in12k_ft_in1k
Image Classification • 32.7M • Updated • 220 • 1 -
timm/mobilenetv4_hybrid_large.ix_e600_r384_in1k
Image Classification • 37.9M • Updated • 270 • 5 -
timm/mobilenetv4_hybrid_large.e600_r384_in1k
Image Classification • 37.9M • Updated • 1.43k • 2
The 20 best models on ImageNet-1k validation set, all pretrained on datasets larger than ImageNet and fine-tuned on ImageNet-1k.
-
timm/eva02_large_patch14_448.mim_m38m_ft_in22k_in1k
Image Classification • 0.3B • Updated • 16.7k • 25 -
timm/eva02_large_patch14_448.mim_in22k_ft_in22k_in1k
Image Classification • 0.3B • Updated • 2.76k • 1 -
timm/eva_giant_patch14_560.m30m_ft_in22k_in1k
Image Classification • 1B • Updated • 2.69k • 3 -
timm/eva02_large_patch14_448.mim_m38m_ft_in1k
Image Classification • 0.3B • Updated • 4.52k • 14
Not the most accurate, but the highest throughput image classification models in timm
-
timm/tinynet_e.in1k
Image Classification • 2.06M • Updated • 3.59k -
timm/mobilenetv3_small_050.lamb_in1k
Image Classification • 1.6M • Updated • 24.3k -
timm/lcnet_050.ra2_in1k
Image Classification • 1.89M • Updated • 6.01k -
timm/mobilenetv3_small_075.lamb_in1k
Image Classification • 2.05M • Updated • 12.1k • 1
timm has a number of unique and exclusive models trained on a 11821 (12k) subset of the full ImageNet-22k
-
timm/convnext_xxlarge.clip_laion2b_soup_ft_in12k
Image Classification • 0.9B • Updated • 133 • 2 -
timm/vit_huge_patch14_clip_224.laion2b_ft_in12k
Image Classification • 0.6B • Updated • 62 • 1 -
timm/vit_large_patch14_clip_224.openai_ft_in12k
Image Classification • 0.3B • Updated • 40 -
timm/vit_large_patch14_clip_224.laion2b_ft_in12k
Image Classification • 0.3B • Updated • 66
timm includes the most popular convolutional and vision transformer models, many with new weights from updated training recipes.
Fastest image classification models with 75.3% accuracy in ImageNet-1k .
-
timm/levit_128s.fb_dist_in1k
Image Classification • 7.82M • Updated • 564 • 2 -
timm/vit_small_patch32_224.augreg_in21k_ft_in1k
Image Classification • 22.9M • Updated • 802 • 2 -
timm/levit_128.fb_dist_in1k
Image Classification • 9.26M • Updated • 6.22k • 1 -
timm/efficientvit_m5.r224_in1k
Image Classification • 12.5M • Updated • 439
Fastest image classification models with 80% accuracy in ImageNet-1k .
-
timm/levit_256.fb_dist_in1k
Image Classification • 19M • Updated • 58k • 2 -
timm/vit_base_patch32_clip_224.laion2b_ft_in1k
Image Classification • 88.2M • Updated • 138 • 1 -
timm/vit_base_patch32_clip_224.laion2b_ft_in12k_in1k
Image Classification • 88.2M • Updated • 1.04k • 4 -
timm/vit_base_patch32_clip_224.openai_ft_in1k
Image Classification • 88.2M • Updated • 503
Fastest image classification models with 83% accuracy in ImageNet-1k .
-
timm/vit_base_patch32_clip_224.laion2b_ft_in12k_in1k
Image Classification • 88.2M • Updated • 1.04k • 4 -
timm/deit3_small_patch16_224.fb_in22k_ft_in1k
Image Classification • 22.1M • Updated • 4.41k -
timm/tiny_vit_11m_224.dist_in22k_ft_in1k
Image Classification • 11M • Updated • 7.2k -
timm/tresnet_m.miil_in21k_ft_in1k
Image Classification • 31.5M • Updated • 255
Fastest image classification models with 86% accuracy in ImageNet-1k .
-
timm/vit_base_patch16_clip_224.laion2b_ft_in12k_in1k
Image Classification • 86.6M • Updated • 4.3k • 2 -
timm/beitv2_base_patch16_224.in1k_ft_in22k_in1k
Image Classification • 87M • Updated • 4.03k -
timm/convnext_base.clip_laion2b_augreg_ft_in12k_in1k
Image Classification • 88.6M • Updated • 107k -
timm/convnext_base.clip_laion2b_augreg_ft_in1k
Image Classification • 88.6M • Updated • 1.27k
Fastest image classification models with 88% accuracy in ImageNet-1k .
-
timm/eva_large_patch14_196.in22k_ft_in22k_in1k
Image Classification • 0.3B • Updated • 12.1k • 3 -
timm/beitv2_large_patch16_224.in1k_ft_in22k_in1k
Image Classification • 0.3B • Updated • 3.03k • 3 -
timm/vit_large_patch14_clip_224.openai_ft_in12k_in1k
Image Classification • 0.3B • Updated • 1.89k • 38 -
timm/convnext_large_mlp.clip_laion2b_soup_ft_in12k_in1k_384
Image Classification • 0.2B • Updated • 5.47k • 4
Pre-trained feature extraction backbones available in timm.
-
timm/vit_small_patch14_dinov2.lvd142m
Image Feature Extraction • 22.1M • Updated • 1.37M • 8 -
timm/vit_large_patch14_dinov2.lvd142m
Image Feature Extraction • 0.3B • Updated • 210k • 17 -
timm/vit_base_patch16_224.dino
Image Feature Extraction • 85.8M • Updated • 75.3k • 6 -
timm/vit_base_patch16_clip_224.openai
Image Feature Extraction • Updated • 156k • 12
Noteworthy instances of ImageNet on the Hub. Vetted and tested with timm train and validation scripts.
Datasets for fine-tune benchmarking, hparam tuning. All vetted and tested with timm scripts.
A collection of very small (~300-500k parameter) models at 160x160 resolution, for testing purposes. Trained on ImageNet-1k.
-
timm/test_byobnet.r160_in1k
Image Classification • 459k • Updated • 1.04k • 1 -
timm/test_convnext.r160_in1k
Image Classification • 272k • Updated • 833 • 1 -
timm/test_convnext2.r160_in1k
Image Classification • 478k • Updated • 877 -
timm/test_convnext3.r160_in1k
Image Classification • 469k • Updated • 831