Instructions to use keypa/MoonViT-V2-Standalone with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use keypa/MoonViT-V2-Standalone with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-feature-extraction", model="keypa/MoonViT-V2-Standalone")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("keypa/MoonViT-V2-Standalone", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Upload vision_config.json with huggingface_hub
Browse files- vision_config.json +30 -0
vision_config.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_attn_implementation": "flash_attention_2",
|
| 3 |
+
"activation_func": "gelu_pytorch_tanh",
|
| 4 |
+
"attn_bias": false,
|
| 5 |
+
"init_pos_emb_height": 64,
|
| 6 |
+
"init_pos_emb_time": 4,
|
| 7 |
+
"init_pos_emb_width": 64,
|
| 8 |
+
"linear_bias": false,
|
| 9 |
+
"merge_kernel_size": [
|
| 10 |
+
2,
|
| 11 |
+
2
|
| 12 |
+
],
|
| 13 |
+
"merge_type": "sd2_tpool",
|
| 14 |
+
"mlp_type": "mlp2",
|
| 15 |
+
"mm_hidden_size": 1024,
|
| 16 |
+
"mm_projector_type": "patchmergerv2",
|
| 17 |
+
"norm_type": "rmsnorm",
|
| 18 |
+
"patch_embed_proj_bias": false,
|
| 19 |
+
"patch_size": 14,
|
| 20 |
+
"pos_emb_interpolation_mode": "bilinear",
|
| 21 |
+
"pos_emb_type": "divided_fixed",
|
| 22 |
+
"projector_hidden_act": "gelu",
|
| 23 |
+
"projector_ln_eps": 1e-05,
|
| 24 |
+
"qkv_hidden_size": 1536,
|
| 25 |
+
"text_hidden_size": 7168,
|
| 26 |
+
"vt_hidden_size": 1024,
|
| 27 |
+
"vt_intermediate_size": 4096,
|
| 28 |
+
"vt_num_attention_heads": 12,
|
| 29 |
+
"vt_num_hidden_layers": 27
|
| 30 |
+
}
|