Image Feature Extraction
Transformers
Safetensors
English
Chinese
mingtok
visual-tokenizer
feature-extraction
image-reconstruction
autoregressive
Instructions to use inclusionAI/MingTok-Vision with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use inclusionAI/MingTok-Vision with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-feature-extraction", model="inclusionAI/MingTok-Vision")# Load model directly from transformers import MingTok model = MingTok.from_pretrained("inclusionAI/MingTok-Vision", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "architectures": [ | |
| "MingTok" | |
| ], | |
| "low_level_encoder": { | |
| "depth": 12, | |
| "embed_dim": 768, | |
| "ffn_layer": "swiglufused", | |
| "img_size": 512, | |
| "out_dim": 32, | |
| "patch_size": 32 | |
| }, | |
| "mean": 1.46817409, | |
| "model_dtype": "bf16", | |
| "model_type": "mingtok", | |
| "pixel_decoder": { | |
| "decoder_depth": 24, | |
| "embed_dim": 1024, | |
| "loss_type": "L1-plain", | |
| "norm_pix_loss": true, | |
| "patch_size": 16 | |
| }, | |
| "pretrained_checkpoint": "/mnt/nativemm-hn/checkpoint/ziyuan/moe_mingtok/mingtok_moe_0830_recon_0830_bef_joint_training_resize_dec_p16d24c1024_s2noresize/gan_v2_20M/202509281648/checkpoint_1_hf.pth", | |
| "scaling_factor": 8.09449291, | |
| "semantic_decoder": { | |
| "decoder_depth": 24, | |
| "embed_dim": 1024, | |
| "ffn_layer": "swiglufused", | |
| "in_dim": 32, | |
| "patch_size": 32 | |
| }, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.52.4" | |
| } | |