Hugging Face's logo

Spaces:
ychenhq
/
VideoCrafterXtend
Runtime error

App Files Community
ychenhq commited on
Commit
abe4aa9
•
1 Parent(s): c62dd62

Upload folder using huggingface_hub

Browse files
Files changed (4) hide show
  1. configs/inference_i2v_512_v1.0.yaml +83 -0
  2. configs/inference_t2v_1024_v1.0.yaml +77 -0
  3. configs/inference_t2v_512_v1.0.yaml +74 -0
  4. configs/inference_t2v_512_v2.0.yaml +77 -0
configs/inference_i2v_512_v1.0.yaml ADDED
@@ -0,0 +1,83 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: lvdm.models.ddpm3d.LatentVisualDiffusion
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.012
6
+ num_timesteps_cond: 1
7
+ timesteps: 1000
8
+ first_stage_key: video
9
+ cond_stage_key: caption
10
+ cond_stage_trainable: false
11
+ conditioning_key: crossattn
12
+ image_size:
13
+ - 40
14
+ - 64
15
+ channels: 4
16
+ scale_by_std: false
17
+ scale_factor: 0.18215
18
+ use_ema: false
19
+ uncond_type: empty_seq
20
+ use_scale: true
21
+ scale_b: 0.7
22
+ finegrained: true
23
+ unet_config:
24
+ target: lvdm.modules.networks.openaimodel3d.UNetModel
25
+ params:
26
+ in_channels: 4
27
+ out_channels: 4
28
+ model_channels: 320
29
+ attention_resolutions:
30
+ - 4
31
+ - 2
32
+ - 1
33
+ num_res_blocks: 2
34
+ channel_mult:
35
+ - 1
36
+ - 2
37
+ - 4
38
+ - 4
39
+ num_head_channels: 64
40
+ transformer_depth: 1
41
+ context_dim: 1024
42
+ use_linear: true
43
+ use_checkpoint: true
44
+ temporal_conv: true
45
+ temporal_attention: true
46
+ temporal_selfatt_only: true
47
+ use_relative_position: false
48
+ use_causal_attention: false
49
+ use_image_attention: true
50
+ temporal_length: 16
51
+ addition_attention: true
52
+ fps_cond: true
53
+ first_stage_config:
54
+ target: lvdm.models.autoencoder.AutoencoderKL
55
+ params:
56
+ embed_dim: 4
57
+ monitor: val/rec_loss
58
+ ddconfig:
59
+ double_z: true
60
+ z_channels: 4
61
+ resolution: 512
62
+ in_channels: 3
63
+ out_ch: 3
64
+ ch: 128
65
+ ch_mult:
66
+ - 1
67
+ - 2
68
+ - 4
69
+ - 4
70
+ num_res_blocks: 2
71
+ attn_resolutions: []
72
+ dropout: 0.0
73
+ lossconfig:
74
+ target: torch.nn.Identity
75
+ cond_stage_config:
76
+ target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
77
+ params:
78
+ freeze: true
79
+ layer: penultimate
80
+ cond_img_config:
81
+ target: lvdm.modules.encoders.condition.FrozenOpenCLIPImageEmbedderV2
82
+ params:
83
+ freeze: true
configs/inference_t2v_1024_v1.0.yaml ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: lvdm.models.ddpm3d.LatentDiffusion
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.012
6
+ num_timesteps_cond: 1
7
+ timesteps: 1000
8
+ first_stage_key: video
9
+ cond_stage_key: caption
10
+ cond_stage_trainable: false
11
+ conditioning_key: crossattn
12
+ image_size:
13
+ - 72
14
+ - 128
15
+ channels: 4
16
+ scale_by_std: false
17
+ scale_factor: 0.18215
18
+ use_ema: false
19
+ uncond_type: empty_seq
20
+ use_scale: true
21
+ fix_scale_bug: true
22
+ unet_config:
23
+ target: lvdm.modules.networks.openaimodel3d.UNetModel
24
+ params:
25
+ in_channels: 4
26
+ out_channels: 4
27
+ model_channels: 320
28
+ attention_resolutions:
29
+ - 4
30
+ - 2
31
+ - 1
32
+ num_res_blocks: 2
33
+ channel_mult:
34
+ - 1
35
+ - 2
36
+ - 4
37
+ - 4
38
+ num_head_channels: 64
39
+ transformer_depth: 1
40
+ context_dim: 1024
41
+ use_linear: true
42
+ use_checkpoint: true
43
+ temporal_conv: false
44
+ temporal_attention: true
45
+ temporal_selfatt_only: true
46
+ use_relative_position: true
47
+ use_causal_attention: false
48
+ temporal_length: 16
49
+ addition_attention: true
50
+ fps_cond: true
51
+ first_stage_config:
52
+ target: lvdm.models.autoencoder.AutoencoderKL
53
+ params:
54
+ embed_dim: 4
55
+ monitor: val/rec_loss
56
+ ddconfig:
57
+ double_z: true
58
+ z_channels: 4
59
+ resolution: 512
60
+ in_channels: 3
61
+ out_ch: 3
62
+ ch: 128
63
+ ch_mult:
64
+ - 1
65
+ - 2
66
+ - 4
67
+ - 4
68
+ num_res_blocks: 2
69
+ attn_resolutions: []
70
+ dropout: 0.0
71
+ lossconfig:
72
+ target: torch.nn.Identity
73
+ cond_stage_config:
74
+ target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
75
+ params:
76
+ freeze: true
77
+ layer: penultimate
configs/inference_t2v_512_v1.0.yaml ADDED
@@ -0,0 +1,74 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: lvdm.models.ddpm3d.LatentDiffusion
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.012
6
+ num_timesteps_cond: 1
7
+ timesteps: 1000
8
+ first_stage_key: video
9
+ cond_stage_key: caption
10
+ cond_stage_trainable: false
11
+ conditioning_key: crossattn
12
+ image_size:
13
+ - 40
14
+ - 64
15
+ channels: 4
16
+ scale_by_std: false
17
+ scale_factor: 0.18215
18
+ use_ema: false
19
+ uncond_type: empty_seq
20
+ unet_config:
21
+ target: lvdm.modules.networks.openaimodel3d.UNetModel
22
+ params:
23
+ in_channels: 4
24
+ out_channels: 4
25
+ model_channels: 320
26
+ attention_resolutions:
27
+ - 4
28
+ - 2
29
+ - 1
30
+ num_res_blocks: 2
31
+ channel_mult:
32
+ - 1
33
+ - 2
34
+ - 4
35
+ - 4
36
+ num_head_channels: 64
37
+ transformer_depth: 1
38
+ context_dim: 1024
39
+ use_linear: true
40
+ use_checkpoint: true
41
+ temporal_conv: false
42
+ temporal_attention: true
43
+ temporal_selfatt_only: true
44
+ use_relative_position: true
45
+ use_causal_attention: false
46
+ temporal_length: 16
47
+ addition_attention: true
48
+ first_stage_config:
49
+ target: lvdm.models.autoencoder.AutoencoderKL
50
+ params:
51
+ embed_dim: 4
52
+ monitor: val/rec_loss
53
+ ddconfig:
54
+ double_z: true
55
+ z_channels: 4
56
+ resolution: 512
57
+ in_channels: 3
58
+ out_ch: 3
59
+ ch: 128
60
+ ch_mult:
61
+ - 1
62
+ - 2
63
+ - 4
64
+ - 4
65
+ num_res_blocks: 2
66
+ attn_resolutions: []
67
+ dropout: 0.0
68
+ lossconfig:
69
+ target: torch.nn.Identity
70
+ cond_stage_config:
71
+ target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
72
+ params:
73
+ freeze: true
74
+ layer: penultimate
configs/inference_t2v_512_v2.0.yaml ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ model:
2
+ target: lvdm.models.ddpm3d.LatentDiffusion
3
+ params:
4
+ linear_start: 0.00085
5
+ linear_end: 0.012
6
+ num_timesteps_cond: 1
7
+ timesteps: 1000
8
+ first_stage_key: video
9
+ cond_stage_key: caption
10
+ cond_stage_trainable: false
11
+ conditioning_key: crossattn
12
+ image_size:
13
+ - 40
14
+ - 64
15
+ channels: 4
16
+ scale_by_std: false
17
+ scale_factor: 0.18215
18
+ use_ema: false
19
+ uncond_type: empty_seq
20
+ use_scale: true
21
+ scale_b: 0.7
22
+ unet_config:
23
+ target: lvdm.modules.networks.openaimodel3d.UNetModel
24
+ params:
25
+ in_channels: 4
26
+ out_channels: 4
27
+ model_channels: 320
28
+ attention_resolutions:
29
+ - 4
30
+ - 2
31
+ - 1
32
+ num_res_blocks: 2
33
+ channel_mult:
34
+ - 1
35
+ - 2
36
+ - 4
37
+ - 4
38
+ num_head_channels: 64
39
+ transformer_depth: 1
40
+ context_dim: 1024
41
+ use_linear: true
42
+ use_checkpoint: true
43
+ temporal_conv: true
44
+ temporal_attention: true
45
+ temporal_selfatt_only: true
46
+ use_relative_position: false
47
+ use_causal_attention: false
48
+ temporal_length: 16
49
+ addition_attention: true
50
+ fps_cond: true
51
+ first_stage_config:
52
+ target: lvdm.models.autoencoder.AutoencoderKL
53
+ params:
54
+ embed_dim: 4
55
+ monitor: val/rec_loss
56
+ ddconfig:
57
+ double_z: true
58
+ z_channels: 4
59
+ resolution: 512
60
+ in_channels: 3
61
+ out_ch: 3
62
+ ch: 128
63
+ ch_mult:
64
+ - 1
65
+ - 2
66
+ - 4
67
+ - 4
68
+ num_res_blocks: 2
69
+ attn_resolutions: []
70
+ dropout: 0.0
71
+ lossconfig:
72
+ target: torch.nn.Identity
73
+ cond_stage_config:
74
+ target: lvdm.modules.encoders.condition.FrozenOpenCLIPEmbedder
75
+ params:
76
+ freeze: true
77
+ layer: penultimate