kendrickfff
commited on
Commit
•
8687b99
1
Parent(s):
a774dae
Update README.md
Browse files
README.md
CHANGED
@@ -64,54 +64,67 @@ Epoch 1/10
|
|
64 |
train Loss: 1.0083 Acc: 0.6850
|
65 |
valid Loss: 0.6304 Acc: 0.7985
|
66 |
Epoch 1 completed in 2109.20 seconds.
|
|
|
67 |
Epoch 2/10
|
68 |
----------
|
69 |
train Loss: 0.7347 Acc: 0.7687
|
70 |
valid Loss: 0.8616 Acc: 0.7307
|
71 |
Epoch 2 completed in 2183.41 seconds.
|
|
|
72 |
Epoch 3/10
|
73 |
----------
|
74 |
train Loss: 0.6510 Acc: 0.7913
|
75 |
valid Loss: 0.5594 Acc: 0.8260
|
76 |
Epoch 3 completed in 2174.55 seconds.
|
|
|
77 |
Epoch 4/10
|
78 |
----------
|
79 |
train Loss: 0.5762 Acc: 0.8126
|
80 |
valid Loss: 0.4006 Acc: 0.8655
|
81 |
Epoch 4 completed in 2166.46 seconds.
|
|
|
82 |
Epoch 5/10
|
83 |
----------
|
84 |
train Loss: 0.5478 Acc: 0.8210
|
85 |
valid Loss: 0.3968 Acc: 0.8793
|
86 |
Epoch 5 completed in 2189.89 seconds.
|
|
|
87 |
Epoch 6/10
|
88 |
----------
|
89 |
train Loss: 0.5223 Acc: 0.8272
|
90 |
valid Loss: 0.4051 Acc: 0.8729
|
91 |
Epoch 6 completed in 2185.71 seconds.
|
|
|
92 |
Epoch 7/10
|
93 |
----------
|
94 |
train Loss: 0.4974 Acc: 0.8355
|
95 |
valid Loss: 0.3223 Acc: 0.9094
|
96 |
Epoch 7 completed in 2184.83 seconds.
|
|
|
97 |
Epoch 8/10
|
98 |
----------
|
99 |
train Loss: 0.3464 Acc: 0.8870
|
100 |
valid Loss: 0.2221 Acc: 0.9338
|
101 |
Epoch 8 completed in 2184.53 seconds.
|
|
|
102 |
Epoch 9/10
|
103 |
----------
|
104 |
train Loss: 0.2896 Acc: 0.9049
|
105 |
valid Loss: 0.2125 Acc: 0.9338
|
106 |
Epoch 9 completed in 2181.82 seconds.
|
|
|
107 |
Epoch 10/10
|
108 |
----------
|
109 |
-
|
110 |
train Loss: 0.2604 Acc: 0.9136
|
|
|
111 |
valid Loss: 0.2076 Acc: 0.9326
|
|
|
112 |
Training complete in 362m 11s
|
|
|
113 |
Best val Acc: 0.9338
|
|
|
114 |
Training Time: Approximately 12 minutes on a single GPU for 10 epochs.
|
|
|
115 |
The model showed high accuracy in predicting common categories such as plastic, paper, and metal, but struggled with classes like shoes and clothes, reflecting the challenges of web-scraped images for such categories.
|
116 |
|
117 |
## Conclusion
|
|
|
64 |
train Loss: 1.0083 Acc: 0.6850
|
65 |
valid Loss: 0.6304 Acc: 0.7985
|
66 |
Epoch 1 completed in 2109.20 seconds.
|
67 |
+
|
68 |
Epoch 2/10
|
69 |
----------
|
70 |
train Loss: 0.7347 Acc: 0.7687
|
71 |
valid Loss: 0.8616 Acc: 0.7307
|
72 |
Epoch 2 completed in 2183.41 seconds.
|
73 |
+
|
74 |
Epoch 3/10
|
75 |
----------
|
76 |
train Loss: 0.6510 Acc: 0.7913
|
77 |
valid Loss: 0.5594 Acc: 0.8260
|
78 |
Epoch 3 completed in 2174.55 seconds.
|
79 |
+
|
80 |
Epoch 4/10
|
81 |
----------
|
82 |
train Loss: 0.5762 Acc: 0.8126
|
83 |
valid Loss: 0.4006 Acc: 0.8655
|
84 |
Epoch 4 completed in 2166.46 seconds.
|
85 |
+
|
86 |
Epoch 5/10
|
87 |
----------
|
88 |
train Loss: 0.5478 Acc: 0.8210
|
89 |
valid Loss: 0.3968 Acc: 0.8793
|
90 |
Epoch 5 completed in 2189.89 seconds.
|
91 |
+
|
92 |
Epoch 6/10
|
93 |
----------
|
94 |
train Loss: 0.5223 Acc: 0.8272
|
95 |
valid Loss: 0.4051 Acc: 0.8729
|
96 |
Epoch 6 completed in 2185.71 seconds.
|
97 |
+
|
98 |
Epoch 7/10
|
99 |
----------
|
100 |
train Loss: 0.4974 Acc: 0.8355
|
101 |
valid Loss: 0.3223 Acc: 0.9094
|
102 |
Epoch 7 completed in 2184.83 seconds.
|
103 |
+
|
104 |
Epoch 8/10
|
105 |
----------
|
106 |
train Loss: 0.3464 Acc: 0.8870
|
107 |
valid Loss: 0.2221 Acc: 0.9338
|
108 |
Epoch 8 completed in 2184.53 seconds.
|
109 |
+
|
110 |
Epoch 9/10
|
111 |
----------
|
112 |
train Loss: 0.2896 Acc: 0.9049
|
113 |
valid Loss: 0.2125 Acc: 0.9338
|
114 |
Epoch 9 completed in 2181.82 seconds.
|
115 |
+
|
116 |
Epoch 10/10
|
117 |
----------
|
|
|
118 |
train Loss: 0.2604 Acc: 0.9136
|
119 |
+
|
120 |
valid Loss: 0.2076 Acc: 0.9326
|
121 |
+
|
122 |
Training complete in 362m 11s
|
123 |
+
|
124 |
Best val Acc: 0.9338
|
125 |
+
|
126 |
Training Time: Approximately 12 minutes on a single GPU for 10 epochs.
|
127 |
+
|
128 |
The model showed high accuracy in predicting common categories such as plastic, paper, and metal, but struggled with classes like shoes and clothes, reflecting the challenges of web-scraped images for such categories.
|
129 |
|
130 |
## Conclusion
|