Upload V-JEPA 2.1 ViT-L MLX (encoder + predictor, RoPE, distilled from ViT-G @384)

e5ddc50 verified 2 months ago

167 Bytes

	{
	"model_name": "vjepa2-vit-l-rope-mlx",
	"components": {
	"encoder": "encoder.safetensors",
	"predictor": "predictor.safetensors"
	},
	"quantized": false
	}