Hugging Face's logo

lvjiameng
/
FAMA-Astro

Feature Extraction
TensorBoard
English
astronomical-image-analysis
masked-autoencoder
self-supervised-learning
computer-vision
vision-transformer
foundation-model
image-classification
object-detection
Model card Files
xet
Metrics Community
lvjiameng commited on
Commit
64319d2
·
verified ·
1 Parent(s): d593c77

Upload 4 files

Browse files
Files changed (4) hide show
  1. base/acc_loss/train_loss.npy +3 -0
  2. base/logs/events.out.tfevents.1724719901.ji-aitrain-8205848960013471744-master-0.91.0 +3 -0
  3. base/model.txt +288 -0
  4. base/weights/best/epoch_777_loss_0.7236/ckpt.pth +3 -0
base/acc_loss/train_loss.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e24f877f084df9383d27b0243504ffc97580626955e01cb91c94e660667d25f1
3
+ size 6528
base/logs/events.out.tfevents.1724719901.ji-aitrain-8205848960013471744-master-0.91.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0edf0711000ae1effdb54b45cd721e8e70f1f3df05c32bfe904d0b388880152
3
+ size 71828
base/model.txt ADDED
@@ -0,0 +1,288 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ MaskedAutoEncoderViT(
2
+ (patch_embed): PatchEmbed(
3
+ (proj): Conv2d(3, 768, kernel_size=(16, 16), stride=(16, 16))
4
+ (norm): Identity()
5
+ )
6
+ (blocks): ModuleList(
7
+ (0): Block(
8
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
9
+ (attn): Attention(
10
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
11
+ (attn_drop): Dropout(p=0.0, inplace=False)
12
+ (proj): Linear(in_features=768, out_features=768, bias=True)
13
+ (proj_drop): Dropout(p=0.0, inplace=False)
14
+ )
15
+ (ls1): Identity()
16
+ (drop_path1): Identity()
17
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
18
+ (mlp): Mlp(
19
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
20
+ (act): GELU(approximate=none)
21
+ (drop1): Dropout(p=0.0, inplace=False)
22
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
23
+ (drop2): Dropout(p=0.0, inplace=False)
24
+ )
25
+ (ls2): Identity()
26
+ (drop_path2): Identity()
27
+ )
28
+ (1): Block(
29
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
30
+ (attn): Attention(
31
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
32
+ (attn_drop): Dropout(p=0.0, inplace=False)
33
+ (proj): Linear(in_features=768, out_features=768, bias=True)
34
+ (proj_drop): Dropout(p=0.0, inplace=False)
35
+ )
36
+ (ls1): Identity()
37
+ (drop_path1): Identity()
38
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
39
+ (mlp): Mlp(
40
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
41
+ (act): GELU(approximate=none)
42
+ (drop1): Dropout(p=0.0, inplace=False)
43
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
44
+ (drop2): Dropout(p=0.0, inplace=False)
45
+ )
46
+ (ls2): Identity()
47
+ (drop_path2): Identity()
48
+ )
49
+ (2): Block(
50
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
51
+ (attn): Attention(
52
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
53
+ (attn_drop): Dropout(p=0.0, inplace=False)
54
+ (proj): Linear(in_features=768, out_features=768, bias=True)
55
+ (proj_drop): Dropout(p=0.0, inplace=False)
56
+ )
57
+ (ls1): Identity()
58
+ (drop_path1): Identity()
59
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
60
+ (mlp): Mlp(
61
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
62
+ (act): GELU(approximate=none)
63
+ (drop1): Dropout(p=0.0, inplace=False)
64
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
65
+ (drop2): Dropout(p=0.0, inplace=False)
66
+ )
67
+ (ls2): Identity()
68
+ (drop_path2): Identity()
69
+ )
70
+ (3): Block(
71
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
72
+ (attn): Attention(
73
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
74
+ (attn_drop): Dropout(p=0.0, inplace=False)
75
+ (proj): Linear(in_features=768, out_features=768, bias=True)
76
+ (proj_drop): Dropout(p=0.0, inplace=False)
77
+ )
78
+ (ls1): Identity()
79
+ (drop_path1): Identity()
80
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
81
+ (mlp): Mlp(
82
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
83
+ (act): GELU(approximate=none)
84
+ (drop1): Dropout(p=0.0, inplace=False)
85
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
86
+ (drop2): Dropout(p=0.0, inplace=False)
87
+ )
88
+ (ls2): Identity()
89
+ (drop_path2): Identity()
90
+ )
91
+ (4): Block(
92
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
93
+ (attn): Attention(
94
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
95
+ (attn_drop): Dropout(p=0.0, inplace=False)
96
+ (proj): Linear(in_features=768, out_features=768, bias=True)
97
+ (proj_drop): Dropout(p=0.0, inplace=False)
98
+ )
99
+ (ls1): Identity()
100
+ (drop_path1): Identity()
101
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
102
+ (mlp): Mlp(
103
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
104
+ (act): GELU(approximate=none)
105
+ (drop1): Dropout(p=0.0, inplace=False)
106
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
107
+ (drop2): Dropout(p=0.0, inplace=False)
108
+ )
109
+ (ls2): Identity()
110
+ (drop_path2): Identity()
111
+ )
112
+ (5): Block(
113
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
114
+ (attn): Attention(
115
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
116
+ (attn_drop): Dropout(p=0.0, inplace=False)
117
+ (proj): Linear(in_features=768, out_features=768, bias=True)
118
+ (proj_drop): Dropout(p=0.0, inplace=False)
119
+ )
120
+ (ls1): Identity()
121
+ (drop_path1): Identity()
122
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
123
+ (mlp): Mlp(
124
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
125
+ (act): GELU(approximate=none)
126
+ (drop1): Dropout(p=0.0, inplace=False)
127
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
128
+ (drop2): Dropout(p=0.0, inplace=False)
129
+ )
130
+ (ls2): Identity()
131
+ (drop_path2): Identity()
132
+ )
133
+ (6): Block(
134
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
135
+ (attn): Attention(
136
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
137
+ (attn_drop): Dropout(p=0.0, inplace=False)
138
+ (proj): Linear(in_features=768, out_features=768, bias=True)
139
+ (proj_drop): Dropout(p=0.0, inplace=False)
140
+ )
141
+ (ls1): Identity()
142
+ (drop_path1): Identity()
143
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
144
+ (mlp): Mlp(
145
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
146
+ (act): GELU(approximate=none)
147
+ (drop1): Dropout(p=0.0, inplace=False)
148
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
149
+ (drop2): Dropout(p=0.0, inplace=False)
150
+ )
151
+ (ls2): Identity()
152
+ (drop_path2): Identity()
153
+ )
154
+ (7): Block(
155
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
156
+ (attn): Attention(
157
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
158
+ (attn_drop): Dropout(p=0.0, inplace=False)
159
+ (proj): Linear(in_features=768, out_features=768, bias=True)
160
+ (proj_drop): Dropout(p=0.0, inplace=False)
161
+ )
162
+ (ls1): Identity()
163
+ (drop_path1): Identity()
164
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
165
+ (mlp): Mlp(
166
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
167
+ (act): GELU(approximate=none)
168
+ (drop1): Dropout(p=0.0, inplace=False)
169
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
170
+ (drop2): Dropout(p=0.0, inplace=False)
171
+ )
172
+ (ls2): Identity()
173
+ (drop_path2): Identity()
174
+ )
175
+ (8): Block(
176
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
177
+ (attn): Attention(
178
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
179
+ (attn_drop): Dropout(p=0.0, inplace=False)
180
+ (proj): Linear(in_features=768, out_features=768, bias=True)
181
+ (proj_drop): Dropout(p=0.0, inplace=False)
182
+ )
183
+ (ls1): Identity()
184
+ (drop_path1): Identity()
185
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
186
+ (mlp): Mlp(
187
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
188
+ (act): GELU(approximate=none)
189
+ (drop1): Dropout(p=0.0, inplace=False)
190
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
191
+ (drop2): Dropout(p=0.0, inplace=False)
192
+ )
193
+ (ls2): Identity()
194
+ (drop_path2): Identity()
195
+ )
196
+ (9): Block(
197
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
198
+ (attn): Attention(
199
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
200
+ (attn_drop): Dropout(p=0.0, inplace=False)
201
+ (proj): Linear(in_features=768, out_features=768, bias=True)
202
+ (proj_drop): Dropout(p=0.0, inplace=False)
203
+ )
204
+ (ls1): Identity()
205
+ (drop_path1): Identity()
206
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
207
+ (mlp): Mlp(
208
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
209
+ (act): GELU(approximate=none)
210
+ (drop1): Dropout(p=0.0, inplace=False)
211
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
212
+ (drop2): Dropout(p=0.0, inplace=False)
213
+ )
214
+ (ls2): Identity()
215
+ (drop_path2): Identity()
216
+ )
217
+ (10): Block(
218
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
219
+ (attn): Attention(
220
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
221
+ (attn_drop): Dropout(p=0.0, inplace=False)
222
+ (proj): Linear(in_features=768, out_features=768, bias=True)
223
+ (proj_drop): Dropout(p=0.0, inplace=False)
224
+ )
225
+ (ls1): Identity()
226
+ (drop_path1): Identity()
227
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
228
+ (mlp): Mlp(
229
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
230
+ (act): GELU(approximate=none)
231
+ (drop1): Dropout(p=0.0, inplace=False)
232
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
233
+ (drop2): Dropout(p=0.0, inplace=False)
234
+ )
235
+ (ls2): Identity()
236
+ (drop_path2): Identity()
237
+ )
238
+ (11): Block(
239
+ (norm1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
240
+ (attn): Attention(
241
+ (qkv): Linear(in_features=768, out_features=2304, bias=True)
242
+ (attn_drop): Dropout(p=0.0, inplace=False)
243
+ (proj): Linear(in_features=768, out_features=768, bias=True)
244
+ (proj_drop): Dropout(p=0.0, inplace=False)
245
+ )
246
+ (ls1): Identity()
247
+ (drop_path1): Identity()
248
+ (norm2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
249
+ (mlp): Mlp(
250
+ (fc1): Linear(in_features=768, out_features=3072, bias=True)
251
+ (act): GELU(approximate=none)
252
+ (drop1): Dropout(p=0.0, inplace=False)
253
+ (fc2): Linear(in_features=3072, out_features=768, bias=True)
254
+ (drop2): Dropout(p=0.0, inplace=False)
255
+ )
256
+ (ls2): Identity()
257
+ (drop_path2): Identity()
258
+ )
259
+ )
260
+ (norm): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
261
+ (decoder_embed): Linear(in_features=768, out_features=512, bias=True)
262
+ (decoder_blocks): ModuleList(
263
+ (0): Block(
264
+ (norm1): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
265
+ (attn): Attention(
266
+ (qkv): Linear(in_features=512, out_features=1536, bias=True)
267
+ (attn_drop): Dropout(p=0.0, inplace=False)
268
+ (proj): Linear(in_features=512, out_features=512, bias=True)
269
+ (proj_drop): Dropout(p=0.0, inplace=False)
270
+ )
271
+ (ls1): Identity()
272
+ (drop_path1): Identity()
273
+ (norm2): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
274
+ (mlp): Mlp(
275
+ (fc1): Linear(in_features=512, out_features=2048, bias=True)
276
+ (act): GELU(approximate=none)
277
+ (drop1): Dropout(p=0.0, inplace=False)
278
+ (fc2): Linear(in_features=2048, out_features=512, bias=True)
279
+ (drop2): Dropout(p=0.0, inplace=False)
280
+ )
281
+ (ls2): Identity()
282
+ (drop_path2): Identity()
283
+ )
284
+ )
285
+ (decoder_norm): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
286
+ (decoder_pred): Linear(in_features=512, out_features=768, bias=True)
287
+ )
288
+ Namespace(batch_size=256, blr=0.0002, dataset='/home/dataset', decoder_depth=1, device='cuda', dist_backend='nccl', dist_url='env://', distributed=True, epochs=800, gpu=0, lr=0.0032, mask_ratio=0.75, min_lr=0.0, model_name='mae_vit_base_patch16', norm_pix_loss=True, rank=0, resume=None, save_dir='./run/ablation/decoder/depth-1', start_epoch=0, sync_bn=True, use_amp=True, warmup_epochs=40, weight_decay=0.05, workers=16, world_size=16)
base/weights/best/epoch_777_loss_0.7236/ckpt.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:78a7e18357ac22a6180ccd46335ea3e33651ee1113b9822ab8266d456149c491
3
+ size 1076290237