LH-Tech-AI commited on
Commit
76a1be0
Β·
verified Β·
1 Parent(s): f1a401c

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +83 -2
README.md CHANGED
@@ -15,11 +15,92 @@ license: apache-2.0
15
  * **Max Input Length:** 20 tokens
16
  * **Performance:** ~97% Accuracy on 4-operation math (Validation Set)
17
 
18
- #### Performance Chart (Current Progress)
19
  | Epoch | Training Loss | Val Accuracy | Status |
20
  | :--- | :--- | :--- | :--- |
21
  | 1 | 1.1420 | 54.89% | πŸ”΄ Learnt Format |
22
  | 2 | 0.3931 | 78.79% | 🟑 Learnt Digits |
23
  | 5 | 0.1638 | 91.91% | 🟒 Learning subtleties |
24
  | 9 | 0.1051 | 97.15% | πŸ”΅ High Precision |
25
- | **10** | **0.1004** | **97.73%** | πŸš€ **Near Perfect** |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
15
  * **Max Input Length:** 20 tokens
16
  * **Performance:** ~97% Accuracy on 4-operation math (Validation Set)
17
 
18
+ ## Performance Chart
19
  | Epoch | Training Loss | Val Accuracy | Status |
20
  | :--- | :--- | :--- | :--- |
21
  | 1 | 1.1420 | 54.89% | πŸ”΄ Learnt Format |
22
  | 2 | 0.3931 | 78.79% | 🟑 Learnt Digits |
23
  | 5 | 0.1638 | 91.91% | 🟒 Learning subtleties |
24
  | 9 | 0.1051 | 97.15% | πŸ”΅ High Precision |
25
+ | **10** | **0.1004** | **97.73%** | πŸš€ **Near Perfect** |
26
+
27
+ ## How to use
28
+ To use this model, download `model.pt` and `use.py` and run it on any type of device with Python3.
29
+
30
+ ## Examples
31
+ Model loaded (Accuracy: 97.73% from epoch 10)
32
+
33
+ --- Mini Math Model interactive ---
34
+ Enter an arithmetic task (e.g. 15*15) or type 'exit' to quit this.
35
+
36
+ Task > 0*567
37
+ Model: 0 | Correct: 0 βœ…
38
+
39
+ Task > 999+999
40
+ Model: 1998 | Correct: 1998 βœ…
41
+
42
+ Task > 1/1
43
+ Model: 1 | Correct: 1 βœ…
44
+
45
+ Task > 1684*8787
46
+ Model: 6398 | Correct: 14797308 ❌
47
+
48
+ Task > 124*598
49
+ Model: 2452 | Correct: 74152 ❌
50
+
51
+ Task > 12/68
52
+ Model: 4 | Correct: 0 ❌
53
+
54
+ Task > 123*123
55
+ Model: 499 | Correct: 15129 ❌
56
+
57
+ Task > 47*5
58
+ Model: 235 | Correct: 235 βœ…
59
+
60
+ Task > 456+125
61
+ Model: 581 | Correct: 581 βœ…
62
+
63
+ Task > 957-234
64
+ Model: 723 | Correct: 723 βœ…
65
+
66
+ Task > 120-7650
67
+ Model: -550 | Correct: -7530 ❌
68
+
69
+ Task > 450-750
70
+ Model: -300 | Correct: -300 βœ…
71
+
72
+ Task > 453-97
73
+ Model: 356 | Correct: 356 βœ…
74
+
75
+ Task > 129-462
76
+ Model: -333 | Correct: -333 βœ…
77
+
78
+ Task > 8*8
79
+ Model: 64 | Correct: 64 βœ…
80
+
81
+ Task > 54*54
82
+ Model: 2916 | Correct: 2916 βœ…
83
+
84
+ Task > 102*78
85
+ Model: 748 | Correct: 7956 ❌
86
+
87
+ Task > 74*9
88
+ Model: 666 | Correct: 666 βœ…
89
+
90
+ Task > 103-34
91
+ Model: 69 | Correct: 69 βœ…
92
+
93
+ ## Overall accuracy
94
+ The overall accuracy after 10 epochs of training is ~97% for tasks with max. 3 digits each like `74*9` or `103-34`.
95
+
96
+ ## Limitations
97
+ The can't do:
98
+ - Tasks with more than 3 digits like `3984-125`
99
+ - Multiplication tasks with numbers above 99 like `293*21`
100
+ - Complex tasks
101
+
102
+ ## Training
103
+ We trained for 10 epochs (~20 minutes of training on Kaggle 2x T4) with 2 million randomly generated training samples.
104
+
105
+ ## Final thoughts
106
+ We may be releasing an improved version of this that can solve really complex tasks and much more...stay tuned!