kaiyuyue commited on
Commit
eecb541
·
verified ·
1 Parent(s): b807850

Add files using upload-large-folder tool

Browse files
sphere-l-af/cfg.json ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "log_interval": 100,
3
+ "vis_interval": 25,
4
+ "ckpt_save_interval": 100,
5
+ "eval_interval": 10,
6
+ "init_from": "scratch",
7
+ "auto_resume": true,
8
+ "use_ema_model": false,
9
+ "dataset_name": "animal-faces",
10
+ "num_classes": 3,
11
+ "image_size": 256,
12
+ "batch_size_per_rank": 8,
13
+ "num_workers": 12,
14
+ "crop_mode": "center",
15
+ "flip_image": true,
16
+ "extra_padding": false,
17
+ "rot_degrees": 0,
18
+ "concat_train_val_splits": true,
19
+ "load_from_zip": true,
20
+ "max_samples": -1,
21
+ "override_model_with_ema": false,
22
+ "batch_size": 256,
23
+ "warmup_epochs": 10,
24
+ "weight_decay": 0.0,
25
+ "grad_clip": 1.0,
26
+ "learning_rate": 0.0001,
27
+ "min_lr": 1.0000000000000002e-06,
28
+ "encoder_lr_scaler": 0.1,
29
+ "epochs": 1000,
30
+ "decay_lr": true,
31
+ "warmup_steps_ratio": 0.01,
32
+ "lr_decay_steps_ratio": 0.99,
33
+ "patch_size": 8,
34
+ "token_channels": 128,
35
+ "vit_enc_model_size": "large",
36
+ "vit_dec_model_size": "large",
37
+ "vit_enc_latent_mlp_mixer_depth": 4,
38
+ "vit_dec_latent_mlp_mixer_depth": 4,
39
+ "affine_latent_mlp_mixer": true,
40
+ "in_context_size": 0,
41
+ "halve_model_size": false,
42
+ "spherify_model": false,
43
+ "cond_generator": false,
44
+ "pixel_head_type": "linear",
45
+ "noise_sigma_max_angle": 85,
46
+ "mix_hard_cases": true,
47
+ "mix_hard_cases_prob": 0.1,
48
+ "mix_hard_cases_max_angle": 89,
49
+ "use_ema": true,
50
+ "ema_model_decay": 0.9997,
51
+ "model_combo": "large-large",
52
+ "device": "cuda",
53
+ "dtype": "bfloat16",
54
+ "compile_model": false
55
+ }
sphere-l-af/ckpt/ep0999.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42cbe1d0700ded86e9de56816691af029f6c3a506873ac55a52af0ff3fae4dbd
3
+ size 2568806154
sphere-l-af/eval/eval_tabl_ep0999_gen_ema=False.txt ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ckpt: ep0999, time: 2026-02-22 11:52:24
3
+ -----
4
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
5
+ |:------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|--------:|-----------:|----------:|----------------:|------:|:---------------|
6
+ | generation | False | False | False | True | extr | 256 | 50000 | 21.5658 | 8.3222 | 0.080676 | 1 | 1 | combo |
7
+
8
+ ckpt: ep0999, time: 2026-02-22 12:20:25
9
+ -----
10
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
11
+ |:------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|-------:|-----------:|----------:|----------------:|------:|:---------------|
12
+ | generation | False | False | False | True | extr | 256 | 50000 | 18.735 | 9.8329 | 0.126945 | 4 | 1 | combo |
sphere-l-imagenet/cfg.json ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "log_interval": 100,
3
+ "vis_interval": 2,
4
+ "ckpt_save_interval": 10,
5
+ "eval_interval": 10,
6
+ "init_from": "scratch",
7
+ "auto_resume": true,
8
+ "dataset_name": "imagenet",
9
+ "num_classes": 1000,
10
+ "class_of_interest": "[913, 283, 17, 954, 429, 416, 751, 625, 607, 933, 723, 911, 629, 599, 148, 992, 207, 360, 387, 974, 88, 979, 417, 279, 966, 717, 10, 928, 953, 936, 866, 107, 574, 486, 393, 269, 967, 815, 950, 644, 852, 130]",
11
+ "image_size": 256,
12
+ "batch_size_per_rank": 8,
13
+ "num_workers": 12,
14
+ "crop_mode": "center",
15
+ "flip_image": true,
16
+ "extra_padding": false,
17
+ "rot_degrees": 0,
18
+ "concat_train_val_splits": false,
19
+ "load_from_zip": false,
20
+ "max_samples": 0,
21
+ "override_with_ema": false,
22
+ "batch_size": 256,
23
+ "warmup_epochs": 5,
24
+ "weight_decay": 0.0,
25
+ "grad_clip": 1.0,
26
+ "learning_rate": 0.0001,
27
+ "min_lr": 1.0000000000000002e-06,
28
+ "encoder_lr_scaler": 0.1,
29
+ "epochs": 800,
30
+ "decay_lr": true,
31
+ "warmup_steps_ratio": 0.00625,
32
+ "lr_decay_steps_ratio": 0.99375,
33
+ "patch_size": 8,
34
+ "token_channels": 64,
35
+ "force_to_bf16": true,
36
+ "vit_enc_model_size": "large",
37
+ "vit_dec_model_size": "large",
38
+ "vit_enc_latent_mlp_mixer_depth": 4,
39
+ "vit_dec_latent_mlp_mixer_depth": 4,
40
+ "affine_latent_mlp_mixer": false,
41
+ "in_context_size": 0,
42
+ "halve_model_size": false,
43
+ "spherify_model": false,
44
+ "cond_generator": true,
45
+ "pixel_head_type": "linear",
46
+ "noise_sigma_max_angle": 85,
47
+ "mix_hard_cases": true,
48
+ "mix_hard_cases_prob": 0.1,
49
+ "mix_hard_cases_max_angle": 89,
50
+ "use_ema": true,
51
+ "ema_model_decay": 0.9997,
52
+ "model_combo": "large-large",
53
+ "device": "cuda",
54
+ "dtype": "bfloat16",
55
+ "compile_model": false
56
+ }
sphere-l-imagenet/ckpt/ep0799.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e2e6d7573b39bcc0d62a2d5813c728561d99a21fda446abd1717d0a9aab1417
3
+ size 1900951002
sphere-l-imagenet/eval/eval_tabl_ep0799_gen_ema=False.txt ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ckpt: ep0799, time: 2026-02-22 12:08:06
3
+ -----
4
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
5
+ |:------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|--------:|-----------:|----------:|----------------:|------:|:---------------|
6
+ | generation | False | False | False | True | rand-50k | 256 | 50000 | 15.6925 | 274.506 | 4.15911 | 1 | 1.4 | combo |
7
+
8
+ ckpt: ep0799, time: 2026-02-22 12:59:43
9
+ -----
10
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
11
+ |:------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|--------:|-----------:|----------:|----------------:|------:|:---------------|
12
+ | generation | False | False | False | True | rand-50k | 256 | 50000 | 4.78979 | 259.145 | 5.43596 | 4 | 1.4 | combo |
sphere-l-imagenet/eval/eval_tabl_ep0799_rec_ema=False.txt ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+
2
+ ckpt: ep0799, time: 2026-02-22 11:55:08
3
+ -----
4
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
5
+ |:---------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|---------:|-----------:|----------:|----------------:|------:|:---------------|
6
+ | reconstruction | False | False | False | False | rand-50k | 256 | 50000 | 0.621119 | 334.426 | 4.73511 | 1 | 1 | combo |
sphere-l-of/cfg.json ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "log_interval": 100,
3
+ "vis_interval": 25,
4
+ "ckpt_save_interval": 100,
5
+ "eval_interval": 10,
6
+ "init_from": "scratch",
7
+ "auto_resume": true,
8
+ "use_ema_model": false,
9
+ "dataset_name": "flowers-102",
10
+ "num_classes": 102,
11
+ "image_size": 256,
12
+ "batch_size_per_rank": 8,
13
+ "num_workers": 12,
14
+ "crop_mode": "center",
15
+ "flip_image": true,
16
+ "extra_padding": false,
17
+ "rot_degrees": 0,
18
+ "concat_train_val_splits": true,
19
+ "load_from_zip": true,
20
+ "max_samples": -1,
21
+ "override_model_with_ema": false,
22
+ "batch_size": 256,
23
+ "warmup_epochs": 10,
24
+ "weight_decay": 0.0,
25
+ "grad_clip": 1.0,
26
+ "learning_rate": 0.0001,
27
+ "min_lr": 1.0000000000000002e-06,
28
+ "encoder_lr_scaler": 0.1,
29
+ "epochs": 1000,
30
+ "decay_lr": true,
31
+ "warmup_steps_ratio": 0.01,
32
+ "lr_decay_steps_ratio": 0.99,
33
+ "patch_size": 8,
34
+ "token_channels": 128,
35
+ "vit_enc_model_size": "large",
36
+ "vit_dec_model_size": "large",
37
+ "vit_enc_latent_mlp_mixer_depth": 4,
38
+ "vit_dec_latent_mlp_mixer_depth": 4,
39
+ "affine_latent_mlp_mixer": true,
40
+ "in_context_size": 0,
41
+ "halve_model_size": false,
42
+ "spherify_model": false,
43
+ "cond_generator": true,
44
+ "pixel_head_type": "linear",
45
+ "noise_sigma_max_angle": 85,
46
+ "mix_hard_cases": true,
47
+ "mix_hard_cases_prob": 0.1,
48
+ "mix_hard_cases_max_angle": 89,
49
+ "use_ema": true,
50
+ "ema_model_decay": 0.9997,
51
+ "model_combo": "large-large",
52
+ "device": "cuda",
53
+ "dtype": "bfloat16",
54
+ "compile_model": false
55
+ }
sphere-l-of/ckpt/ep0999.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d97ec7be8ceb79516aec3e8293949ae4a923b8094ccfa35c32c5cad5cfbb46bc
3
+ size 3795623700
sphere-l-of/eval/eval_tabl_ep0999_gen_ema=False.txt ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ckpt: ep0999, time: 2026-02-22 19:55:21
2
+ -----
3
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
4
+ |:------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|--------:|-----------:|----------:|----------------:|------:|:---------------|
5
+ | generation | False | False | False | False | extr | 256 | 51000 | 25.1064 | 3.41502 | 0.0317006 | 1 | 1.6 | combo |
6
+
7
+ ckpt: ep0999, time: 2026-02-22 21:46:50
8
+ -----
9
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
10
+ |:------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|--------:|-----------:|----------:|----------------:|------:|:---------------|
11
+ | generation | False | False | False | False | extr | 256 | 51000 | 11.2701 | 3.22581 | 0.0372179 | 4 | 1.6 | combo |
sphere-xl-imagenet/cfg.json ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "log_interval": 100,
3
+ "vis_interval": 2,
4
+ "ckpt_save_interval": 5,
5
+ "eval_interval": 10,
6
+ "init_from": "scratch",
7
+ "auto_resume": true,
8
+ "dataset_name": "imagenet",
9
+ "num_classes": 1000,
10
+ "class_of_interest": "[913, 283, 17, 954, 429, 416, 751, 625, 607, 933, 723, 911, 629, 599, 148, 992, 207, 360, 387, 974, 88, 979, 417, 279, 966, 717, 10, 928, 953, 936, 866, 107, 574, 486, 393, 269, 967, 815, 950, 644, 852, 130]",
11
+ "image_size": 256,
12
+ "batch_size_per_rank": 8,
13
+ "num_workers": 12,
14
+ "crop_mode": "center",
15
+ "flip_image": true,
16
+ "extra_padding": false,
17
+ "rot_degrees": 0,
18
+ "concat_train_val_splits": false,
19
+ "load_from_zip": false,
20
+ "max_samples": 0,
21
+ "override_model_with_ema": false,
22
+ "batch_size": 256,
23
+ "warmup_epochs": 5,
24
+ "weight_decay": 0.0,
25
+ "grad_clip": 1.0,
26
+ "learning_rate": 0.0001,
27
+ "min_lr": 1.0000000000000002e-06,
28
+ "encoder_lr_scaler": 0.1,
29
+ "epochs": 800,
30
+ "decay_lr": true,
31
+ "warmup_steps_ratio": 0.00625,
32
+ "lr_decay_steps_ratio": 0.99375,
33
+ "patch_size": 8,
34
+ "token_channels": 64,
35
+ "force_to_bf16": true,
36
+ "vit_enc_model_size": "xlarge",
37
+ "vit_dec_model_size": "xlarge",
38
+ "vit_enc_latent_mlp_mixer_depth": 4,
39
+ "vit_dec_latent_mlp_mixer_depth": 4,
40
+ "affine_latent_mlp_mixer": false,
41
+ "in_context_size": 0,
42
+ "halve_model_size": false,
43
+ "spherify_model": false,
44
+ "cond_generator": true,
45
+ "pixel_head_type": "linear",
46
+ "noise_sigma_max_angle": 85,
47
+ "mix_hard_cases": true,
48
+ "mix_hard_cases_prob": 0.1,
49
+ "mix_hard_cases_max_angle": 89,
50
+ "use_ema": true,
51
+ "ema_model_decay": 0.9997,
52
+ "model_combo": "xlarge-xlarge",
53
+ "device": "cuda",
54
+ "dtype": "bfloat16",
55
+ "compile_model": false
56
+ }
sphere-xl-imagenet/ckpt/ep0799.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61226fdd61d67004aac649935e72a06391f000df18a0ddde6af587d1d5c8a63d
3
+ size 2778122362
sphere-xl-imagenet/eval/eval_tabl_ep0799_gen_ema=False.txt ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ckpt: ep0799, time: 2026-02-22 11:54:36
3
+ -----
4
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
5
+ |:------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|--------:|-----------:|----------:|----------------:|------:|:---------------|
6
+ | generation | False | False | False | True | rand-50k | 256 | 50000 | 14.5296 | 299.397 | 4.4479 | 1 | 1.4 | combo |
7
+
8
+ ckpt: ep0799, time: 2026-02-22 12:40:34
9
+ -----
10
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
11
+ |:------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|--------:|-----------:|----------:|----------------:|------:|:---------------|
12
+ | generation | False | False | False | True | rand-50k | 256 | 50000 | 4.05843 | 266.004 | 3.2567 | 4 | 1.4 | combo |
sphere-xl-imagenet/eval/eval_tabl_ep0799_rec_ema=False.txt ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+
2
+ ckpt: ep0799, time: 2026-02-22 11:41:39
3
+ -----
4
+ | task_mode | use_ema | seed_sampling | use_sampling_scheduler | cache_sampling_noise | fid_stats_used_from | image_size | num_imgs | fid | isc_mean | isc_std | forward_steps | cfg | cfg_position |
5
+ |:---------------|:----------|:----------------|:-------------------------|:-----------------------|:----------------------|-------------:|-----------:|---------:|-----------:|----------:|----------------:|------:|:---------------|
6
+ | reconstruction | False | False | False | False | rand-50k | 256 | 50000 | 0.623909 | 335.129 | 4.4706 | 1 | 1 | combo |