readme: add model card
Browse files
README.md
CHANGED
|
@@ -1,3 +1,107 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
Anthropic's client side tokenizer.
|
| 6 |
+
|
| 7 |
+
Accuracy compared to actual Claude 3 Haiku tokenizer (Claude 3 family has the same tokenizer):
|
| 8 |
+
|
| 9 |
+
```python
|
| 10 |
+
Tokenization results saved to __temp.txt.tokens
|
| 11 |
+
Text: Hello, world! This is a simple...
|
| 12 |
+
Actual tokens: 17
|
| 13 |
+
Predicted tokens: 10
|
| 14 |
+
Accuracy: 58.82%
|
| 15 |
+
--------------------------------------------------
|
| 16 |
+
Tokenization results saved to __temp.txt.tokens
|
| 17 |
+
Text: The quick brown fox jumps over...
|
| 18 |
+
Actual tokens: 19
|
| 19 |
+
Predicted tokens: 10
|
| 20 |
+
Accuracy: 52.63%
|
| 21 |
+
--------------------------------------------------
|
| 22 |
+
Tokenization results saved to __temp.txt.tokens
|
| 23 |
+
Text: In computer programming, a hel...
|
| 24 |
+
Actual tokens: 29
|
| 25 |
+
Predicted tokens: 21
|
| 26 |
+
Accuracy: 72.41%
|
| 27 |
+
--------------------------------------------------
|
| 28 |
+
Tokenization results saved to __temp.txt.tokens
|
| 29 |
+
Text: Artificial intelligence (AI) i...
|
| 30 |
+
Actual tokens: 30
|
| 31 |
+
Predicted tokens: 24
|
| 32 |
+
Accuracy: 80.00%
|
| 33 |
+
--------------------------------------------------
|
| 34 |
+
Tokenization results saved to __temp.txt.tokens
|
| 35 |
+
Text: The Eiffel Tower is a wrought-...
|
| 36 |
+
Actual tokens: 56
|
| 37 |
+
Predicted tokens: 48
|
| 38 |
+
Accuracy: 85.71%
|
| 39 |
+
--------------------------------------------------
|
| 40 |
+
Tokenization results saved to __temp.txt.tokens
|
| 41 |
+
Text: To be, or not to be, that is t...
|
| 42 |
+
Actual tokens: 60
|
| 43 |
+
Predicted tokens: 50
|
| 44 |
+
Accuracy: 83.33%
|
| 45 |
+
--------------------------------------------------
|
| 46 |
+
Tokenization results saved to __temp.txt.tokens
|
| 47 |
+
Text: In the beginning God created t...
|
| 48 |
+
Actual tokens: 38
|
| 49 |
+
Predicted tokens: 31
|
| 50 |
+
Accuracy: 81.58%
|
| 51 |
+
--------------------------------------------------
|
| 52 |
+
Tokenization results saved to __temp.txt.tokens
|
| 53 |
+
Text: Four score and seven years ago...
|
| 54 |
+
Actual tokens: 41
|
| 55 |
+
Predicted tokens: 34
|
| 56 |
+
Accuracy: 82.93%
|
| 57 |
+
--------------------------------------------------
|
| 58 |
+
Tokenization results saved to __temp.txt.tokens
|
| 59 |
+
Text: I have a dream that one day th...
|
| 60 |
+
Actual tokens: 51
|
| 61 |
+
Predicted tokens: 43
|
| 62 |
+
Accuracy: 84.31%
|
| 63 |
+
--------------------------------------------------
|
| 64 |
+
Tokenization results saved to __temp.txt.tokens
|
| 65 |
+
Text: That's one small step for man,...
|
| 66 |
+
Actual tokens: 22
|
| 67 |
+
Predicted tokens: 14
|
| 68 |
+
Accuracy: 63.64%
|
| 69 |
+
--------------------------------------------------
|
| 70 |
+
Tokenization results saved to __temp.txt.tokens
|
| 71 |
+
Text: Here are the key points about ...
|
| 72 |
+
Actual tokens: 203
|
| 73 |
+
Predicted tokens: 195
|
| 74 |
+
Accuracy: 96.06%
|
| 75 |
+
--------------------------------------------------
|
| 76 |
+
Tokenization results saved to __temp.txt.tokens
|
| 77 |
+
Text: This appears to be an excerpt ...
|
| 78 |
+
Actual tokens: 179
|
| 79 |
+
Predicted tokens: 180
|
| 80 |
+
Accuracy: 99.44%
|
| 81 |
+
--------------------------------------------------
|
| 82 |
+
Tokenization results saved to __temp.txt.tokens
|
| 83 |
+
Text: This is the beginning of the b...
|
| 84 |
+
Actual tokens: 194
|
| 85 |
+
Predicted tokens: 191
|
| 86 |
+
Accuracy: 98.45%
|
| 87 |
+
--------------------------------------------------
|
| 88 |
+
Tokenization results saved to __temp.txt.tokens
|
| 89 |
+
Text: That is the opening lines of t...
|
| 90 |
+
Actual tokens: 177
|
| 91 |
+
Predicted tokens: 163
|
| 92 |
+
Accuracy: 92.09%
|
| 93 |
+
--------------------------------------------------
|
| 94 |
+
Tokenization results saved to __temp.txt.tokens
|
| 95 |
+
Text: That's a powerful and inspirin...
|
| 96 |
+
Actual tokens: 193
|
| 97 |
+
Predicted tokens: 190
|
| 98 |
+
Accuracy: 98.45%
|
| 99 |
+
--------------------------------------------------
|
| 100 |
+
Tokenization results saved to __temp.txt.tokens
|
| 101 |
+
Text: That famous quote is from Neil...
|
| 102 |
+
Actual tokens: 131
|
| 103 |
+
Predicted tokens: 122
|
| 104 |
+
Accuracy: 93.13%
|
| 105 |
+
--------------------------------------------------
|
| 106 |
+
Average accuracy: 82.69%
|
| 107 |
+
```
|