Aller au contenu
ModelPrésentéExterne
Télécharger

V JEPA2 Vitl Fpc16 256 Ssv2

Apple MLX fp16 port of facebook/vjepa2-vitl-fpc16-256-ssv2 — V-JEPA2 ViT-L encoder + attentive-pooler + linear head for Something-Something-v2 action recognition (174 classes). MIT.

Sourcehuggingface.co/mlx-community/V-JEPA2-vitl-fpc16-256-ssv2/tree/3669d14505b272517898b2df282f499a2bd97bcf
TypeOtherVersionRevision 3669d14BClasse

Source du modèle

Type
Other
Taille du fichier
674 MB
Version
vRevision 3669d14
Mis à jour
11 oct. 2026
Téléchargements
15
Source
Hugging Face

Description de la source

V-JEPA2 ViT-L SSv2 classifier (MLX)

Apple MLX fp16 port of facebook/vjepa2-vitl-fpc16-256-ssv2 — V-JEPA2 ViT-L encoder + attentive-pooler + linear head for Something-Something-v2 action recognition (174 classes). MIT.

pip install vjepa2-mlx   # https://github.com/xocialize/vjepa2-mlx
vjepa2-mlx -i clip.mp4 --task classify
  • Parity vs PyTorch (cpu fp32): logits rel 2.43e-6, argmax matches (fp16 too).
  • Precision: fp16 (~707 MB).

MIT (© Meta Platforms).

Sources

1 source
Hugging FacePrincipalVersion 3669d14505b272517898b2df282f499a2bd97bcf
huggingface.co/mlx-community/V-JEPA2-vitl-fpc16-256-ssv2/tree/3669d14505b272517898b2df282f499a2bd97bcf

Vérifié 11 oct.

Artefacts du modèle

1 artefact

model.safetensors

safetensors · 674 MB · SHA-256 453b7bf9b1c9…4a40 · Hugging Face

Download file

Extraits de sources

3 extraits
Readme2 aoûthuggingface.co/mlx-community/V-JEPA2-vitl-fpc16-256-ssv2

--- license: mit library_name: mlx pipeline_tag: video-classification tags: [mlx, video, vjepa2, video-classification, something-something-v2, apple-silicon] --- # V-JEPA2 ViT-L SSv2 classifier (MLX) Apple **MLX** fp16 port of `facebook/vjepa2-vitl-fpc16-256-ssv2` — V-JEPA2 ViT-L encoder + attentive-pooler + linear head for **Something-Something-v2** action recognition (174 classes). MIT. ```bash pip install vjepa2-mlx # https://github.com/xocialize/vjepa2-mlx vjepa2-mlx -i clip.mp4 --task classify ``` - **Parity vs PyTorch (cpu fp32)**: logits rel 2.43e-6, argmax matches (fp16 too). - **Precision**: fp16 (~707 MB). MIT (© Meta Platforms).

Note11 oct.huggingface.co/mlx-community/V-JEPA2-vitl-fpc16-256-ssv2

Source context: 15 downloads · 1 likes · Pipeline video-classification · Library mlx · Repo mlx-community/V-JEPA2-vitl-fpc16-256-ssv2

Note2 aoûthuggingface.co/mlx-community/V-JEPA2-vitl-fpc16-256-ssv2

Source context: 18 downloads · 1 likes · Pipeline video-classification · Library mlx · Repo mlx-community/V-JEPA2-vitl-fpc16-256-ssv2