Hugging Face's logo

Vrjb
/
BLIP_Captioning

Image-to-Text
Transformers
Safetensors
blip
Generated from Trainer
Model card Files
xet
Community
Vrjb commited on
Commit
db42bb3
·
verified ·
1 Parent(s): 2d299c4

Model save

Browse files
Files changed (3) hide show
  1. README.md +15 -118
  2. model.safetensors +1 -1
  3. training_args.bin +1 -1
README.md CHANGED
@@ -16,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  This model is a fine-tuned version of [Salesforce/blip-image-captioning-base](https://huggingface.co/Salesforce/blip-image-captioning-base) on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
- - Loss: 0.0147
20
 
21
  ## Model description
22
 
@@ -36,134 +36,31 @@ More information needed
36
 
37
  The following hyperparameters were used during training:
38
  - learning_rate: 0.0001
39
- - train_batch_size: 2
40
  - eval_batch_size: 1
41
  - seed: 42
42
  - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
43
  - lr_scheduler_type: linear
44
  - lr_scheduler_warmup_steps: 1000
45
- - num_epochs: 5
 
46
 
47
  ### Training results
48
 
49
- | Training Loss | Epoch | Step | Validation Loss |
50
- |:-------------:|:------:|:-----:|:---------------:|
51
- | 0.047 | 0.0446 | 500 | 0.0336 |
52
- | 0.0379 | 0.0892 | 1000 | 0.0419 |
53
- | 0.0244 | 0.1339 | 1500 | 0.0247 |
54
- | 0.0279 | 0.1785 | 2000 | 0.0262 |
55
- | 0.0359 | 0.2231 | 2500 | 0.0326 |
56
- | 0.0263 | 0.2677 | 3000 | 0.0245 |
57
- | 0.024 | 0.3124 | 3500 | 0.0239 |
58
- | 0.0247 | 0.3570 | 4000 | 0.0312 |
59
- | 0.0476 | 0.4016 | 4500 | 0.0455 |
60
- | 0.0278 | 0.4462 | 5000 | 0.0243 |
61
- | 0.0231 | 0.4909 | 5500 | 0.0232 |
62
- | 0.0233 | 0.5355 | 6000 | 0.0230 |
63
- | 0.0238 | 0.5801 | 6500 | 0.0236 |
64
- | 0.0227 | 0.6247 | 7000 | 0.0231 |
65
- | 0.0259 | 0.6693 | 7500 | 0.0233 |
66
- | 0.0304 | 0.7140 | 8000 | 0.0273 |
67
- | 0.0225 | 0.7586 | 8500 | 0.0238 |
68
- | 0.0228 | 0.8032 | 9000 | 0.0232 |
69
- | 0.0235 | 0.8478 | 9500 | 0.0282 |
70
- | 0.0219 | 0.8925 | 10000 | 0.0241 |
71
- | 0.0221 | 0.9371 | 10500 | 0.0228 |
72
- | 0.0229 | 0.9817 | 11000 | 0.0229 |
73
- | 0.0227 | 1.0263 | 11500 | 0.0223 |
74
- | 0.0229 | 1.0710 | 12000 | 0.0222 |
75
- | 0.022 | 1.1156 | 12500 | 0.0213 |
76
- | 0.022 | 1.1602 | 13000 | 0.0230 |
77
- | 0.0221 | 1.2048 | 13500 | 0.0253 |
78
- | 0.0219 | 1.2494 | 14000 | 0.0213 |
79
- | 0.0218 | 1.2941 | 14500 | 0.0207 |
80
- | 0.0219 | 1.3387 | 15000 | 0.0214 |
81
- | 0.0221 | 1.3833 | 15500 | 0.0297 |
82
- | 0.0252 | 1.4279 | 16000 | 0.0214 |
83
- | 0.0221 | 1.4726 | 16500 | 0.0206 |
84
- | 0.0218 | 1.5172 | 17000 | 0.0205 |
85
- | 0.0217 | 1.5618 | 17500 | 0.0205 |
86
- | 0.0208 | 1.6064 | 18000 | 0.0213 |
87
- | 0.0222 | 1.6510 | 18500 | 0.0207 |
88
- | 0.0212 | 1.6957 | 19000 | 0.0214 |
89
- | 0.021 | 1.7403 | 19500 | 0.0205 |
90
- | 0.0221 | 1.7849 | 20000 | 0.0205 |
91
- | 0.0213 | 1.8295 | 20500 | 0.0219 |
92
- | 0.0204 | 1.8742 | 21000 | 0.0202 |
93
- | 0.0213 | 1.9188 | 21500 | 0.0204 |
94
- | 0.02 | 1.9634 | 22000 | 0.0204 |
95
- | 0.0217 | 2.0080 | 22500 | 0.0206 |
96
- | 0.0243 | 2.0527 | 23000 | 0.0232 |
97
- | 0.0209 | 2.0973 | 23500 | 0.0215 |
98
- | 0.0209 | 2.1419 | 24000 | 0.0200 |
99
- | 0.0206 | 2.1865 | 24500 | 0.0199 |
100
- | 0.0209 | 2.2311 | 25000 | 0.0197 |
101
- | 0.0204 | 2.2758 | 25500 | 0.0198 |
102
- | 0.0207 | 2.3204 | 26000 | 0.0197 |
103
- | 0.0203 | 2.3650 | 26500 | 0.0196 |
104
- | 0.0201 | 2.4096 | 27000 | 0.0200 |
105
- | 0.0199 | 2.4543 | 27500 | 0.0195 |
106
- | 0.0203 | 2.4989 | 28000 | 0.0199 |
107
- | 0.0199 | 2.5435 | 28500 | 0.0193 |
108
- | 0.0302 | 2.5881 | 29000 | 0.0257 |
109
- | 0.0197 | 2.6328 | 29500 | 0.0197 |
110
- | 0.0197 | 2.6774 | 30000 | 0.0197 |
111
- | 0.02 | 2.7220 | 30500 | 0.0193 |
112
- | 0.0197 | 2.7666 | 31000 | 0.0191 |
113
- | 0.0185 | 2.8112 | 31500 | 0.0192 |
114
- | 0.037 | 2.8559 | 32000 | 0.0258 |
115
- | 0.019 | 2.9005 | 32500 | 0.0192 |
116
- | 0.0194 | 2.9451 | 33000 | 0.0187 |
117
- | 0.0199 | 2.9897 | 33500 | 0.0189 |
118
- | 0.0187 | 3.0344 | 34000 | 0.0187 |
119
- | 0.0192 | 3.0790 | 34500 | 0.0185 |
120
- | 0.0185 | 3.1236 | 35000 | 0.0184 |
121
- | 0.02 | 3.1682 | 35500 | 0.0184 |
122
- | 0.0189 | 3.2129 | 36000 | 0.0183 |
123
- | 0.0196 | 3.2575 | 36500 | 0.0181 |
124
- | 0.0208 | 3.3021 | 37000 | 0.0200 |
125
- | 0.0193 | 3.3467 | 37500 | 0.0183 |
126
- | 0.019 | 3.3913 | 38000 | 0.0180 |
127
- | 0.0181 | 3.4360 | 38500 | 0.0178 |
128
- | 0.0183 | 3.4806 | 39000 | 0.0181 |
129
- | 0.0181 | 3.5252 | 39500 | 0.0177 |
130
- | 0.0172 | 3.5698 | 40000 | 0.0176 |
131
- | 0.0179 | 3.6145 | 40500 | 0.0177 |
132
- | 0.0175 | 3.6591 | 41000 | 0.0174 |
133
- | 0.0171 | 3.7037 | 41500 | 0.0174 |
134
- | 0.02 | 3.7483 | 42000 | 0.0183 |
135
- | 0.0173 | 3.7929 | 42500 | 0.0170 |
136
- | 0.0165 | 3.8376 | 43000 | 0.0167 |
137
- | 0.0189 | 3.8822 | 43500 | 0.0168 |
138
- | 0.0167 | 3.9268 | 44000 | 0.0170 |
139
- | 0.0171 | 3.9714 | 44500 | 0.0164 |
140
- | 0.0172 | 4.0161 | 45000 | 0.0163 |
141
- | 0.0164 | 4.0607 | 45500 | 0.0163 |
142
- | 0.0154 | 4.1053 | 46000 | 0.0161 |
143
- | 0.0163 | 4.1499 | 46500 | 0.0161 |
144
- | 0.0169 | 4.1946 | 47000 | 0.0160 |
145
- | 0.0151 | 4.2392 | 47500 | 0.0165 |
146
- | 0.0152 | 4.2838 | 48000 | 0.0157 |
147
- | 0.015 | 4.3284 | 48500 | 0.0163 |
148
- | 0.0155 | 4.3730 | 49000 | 0.0154 |
149
- | 0.016 | 4.4177 | 49500 | 0.0154 |
150
- | 0.0162 | 4.4623 | 50000 | 0.0153 |
151
- | 0.0157 | 4.5069 | 50500 | 0.0154 |
152
- | 0.0159 | 4.5515 | 51000 | 0.0151 |
153
- | 0.0156 | 4.5962 | 51500 | 0.0155 |
154
- | 0.0147 | 4.6408 | 52000 | 0.0153 |
155
- | 0.0164 | 4.6854 | 52500 | 0.0158 |
156
- | 0.0136 | 4.7300 | 53000 | 0.0160 |
157
- | 0.0146 | 4.7747 | 53500 | 0.0150 |
158
- | 0.0143 | 4.8193 | 54000 | 0.0152 |
159
- | 0.0153 | 4.8639 | 54500 | 0.0149 |
160
- | 0.0145 | 4.9085 | 55000 | 0.0148 |
161
- | 0.0139 | 4.9531 | 55500 | 0.0148 |
162
- | 0.0138 | 4.9978 | 56000 | 0.0147 |
163
 
164
 
165
  ### Framework versions
166
 
167
  - Transformers 4.55.4
168
- - Pytorch 2.1.2+cu121
169
  - Tokenizers 0.21.4
 
16
 
17
  This model is a fine-tuned version of [Salesforce/blip-image-captioning-base](https://huggingface.co/Salesforce/blip-image-captioning-base) on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
+ - Loss: 0.0001
20
 
21
  ## Model description
22
 
 
36
 
37
  The following hyperparameters were used during training:
38
  - learning_rate: 0.0001
39
+ - train_batch_size: 16
40
  - eval_batch_size: 1
41
  - seed: 42
42
  - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
43
  - lr_scheduler_type: linear
44
  - lr_scheduler_warmup_steps: 1000
45
+ - num_epochs: 3
46
+ - mixed_precision_training: Native AMP
47
 
48
  ### Training results
49
 
50
+ | Training Loss | Epoch | Step | Validation Loss |
51
+ |:-------------:|:------:|:----:|:---------------:|
52
+ | 0.0083 | 0.3569 | 500 | 0.0039 |
53
+ | 0.0055 | 0.7138 | 1000 | 0.0386 |
54
+ | 0.0025 | 1.0707 | 1500 | 0.0007 |
55
+ | 0.0016 | 1.4276 | 2000 | 0.0003 |
56
+ | 0.001 | 1.7844 | 2500 | 0.0003 |
57
+ | 0.0007 | 2.1413 | 3000 | 0.0002 |
58
+ | 0.0005 | 2.4982 | 3500 | 0.0001 |
59
+ | 0.0003 | 2.8551 | 4000 | 0.0001 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
60
 
61
 
62
  ### Framework versions
63
 
64
  - Transformers 4.55.4
65
+ - Pytorch 2.5.1+cu121
66
  - Tokenizers 0.21.4
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2ebec43d4beb4014c85524d897d0992fc8aceb8392a306a18b9c6877ad6b6760
3
  size 989717056
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b5bd17c53cae8c47b32643329ae0d4b448c500f8dbdb47fd7e492bb14d7e22ba
3
  size 989717056
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:88f3f784fafea1e6968ecf39355382b5ffb001810d87f8cdd5a4de43739a255b
3
  size 5560
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a37f9c9017ee5c165d48f0d297bd5b323d62c141b346805ea47bc9dd38ef6a7
3
  size 5560