StreamFormer commited on
Commit
a2fbe0c
·
verified ·
1 Parent(s): 6f0640b

Upload folder using huggingface_hub

Browse files
Files changed (2) hide show
  1. config.json +114 -0
  2. model.safetensors +3 -0
config.json ADDED
@@ -0,0 +1,114 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "VFMMultiFrameTransformer"
4
+ ],
5
+ "chosen_layers": [
6
+ 4,
7
+ 11,
8
+ 17,
9
+ 23
10
+ ],
11
+ "ffn_layer": "mlp",
12
+ "geometry_aggregator": false,
13
+ "geometry_aggregator_layer": 6,
14
+ "grounding_ratio": 0.5,
15
+ "hidden_act": "gelu",
16
+ "hidden_size": 1024,
17
+ "image_aggregator": false,
18
+ "image_aggregator_layer": 6,
19
+ "image_size": 224,
20
+ "image_ssl": {
21
+ "compute_precision": {
22
+ "sharding_strategy": "SHARD_GRAD_OP"
23
+ },
24
+ "crops": {
25
+ "local_crops_number": 2
26
+ },
27
+ "dino": {
28
+ "force_weight_norm": false,
29
+ "global_ignore_diagonal": true,
30
+ "head_bottleneck_dim": 256,
31
+ "head_hidden_dim": 2048,
32
+ "head_n_prototypes": 65536,
33
+ "head_nlayers": 3,
34
+ "head_norm_last_layer": false,
35
+ "koleo_distributed_replicas": 0,
36
+ "koleo_loss_distributed": false,
37
+ "koleo_loss_weight": 0.1,
38
+ "koleo_topk": 1,
39
+ "local_loss_weight_schedule": {
40
+ "end": 0.5,
41
+ "peak": 0.5,
42
+ "start": 0.5,
43
+ "warmup_epochs": 0
44
+ },
45
+ "loss_weight": 1.0,
46
+ "reweight_dino_local_loss": false
47
+ },
48
+ "distillation": {
49
+ "checkpoint_path": "",
50
+ "enabled": false,
51
+ "full_cfg_path": ""
52
+ },
53
+ "gram": {
54
+ "ckpt": null,
55
+ "compute_stats": false,
56
+ "ema_teacher": false,
57
+ "global_teacher_resize_antialias": false,
58
+ "global_teacher_resize_method": "bicubic",
59
+ "img_level": true,
60
+ "it_first_update": 0,
61
+ "it_load_ema_teacher": -1,
62
+ "loss_weight": 1.0,
63
+ "loss_weight_schedule": null,
64
+ "max_updates": null,
65
+ "normalized": true,
66
+ "remove_neg": false,
67
+ "remove_only_teacher_neg": false,
68
+ "rep_update": true,
69
+ "tokens_used": "all",
70
+ "update_frequency": 50000,
71
+ "use_loss": true
72
+ },
73
+ "ibot": {
74
+ "force_masking_even_with_zero_weight": false,
75
+ "head_bottleneck_dim": 256,
76
+ "head_hidden_dim": 2048,
77
+ "head_n_prototypes": 65536,
78
+ "head_nlayers": 3,
79
+ "head_norm_last_layer": false,
80
+ "loss_weight": 1.0,
81
+ "mask_random_circular_shift": false,
82
+ "mask_ratio_min_max": [
83
+ 0.1,
84
+ 0.5
85
+ ],
86
+ "mask_sample_probability": 0.5,
87
+ "separate_head": true
88
+ },
89
+ "multidistillation": {
90
+ "enabled": false
91
+ },
92
+ "train": {
93
+ "centering": "sinkhorn_knopp"
94
+ }
95
+ },
96
+ "initializer_range": 0.02,
97
+ "intermediate_size": 3072,
98
+ "layer_norm_eps": 1e-06,
99
+ "mlp_ratio": 4.0,
100
+ "mm_projector_type": "mlp2x_gelu",
101
+ "model_type": "vfm",
102
+ "num_attention_heads": 16,
103
+ "num_channels": 3,
104
+ "num_experts": 8,
105
+ "num_frames": 16,
106
+ "patch_embed_name": "dinov3_vitl16_torch",
107
+ "patch_size": 16,
108
+ "top_k": 2,
109
+ "torch_dtype": "float32",
110
+ "transformers_version": "4.52.3",
111
+ "upcycle_to_moe": false,
112
+ "video_aggregator": true,
113
+ "video_aggregator_layer": 24
114
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9da399c61360115d956466fd1dda29f457ae4a37ffb0ff9b3a9a357bc0033d2
3
+ size 1212686304