HackMD - Collaborative Markdown Knowledge Base

## UttutPipe ## Forward ### Input RAW DATA - label = [0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 2, 2, 3, 3, 3, 0, 0, 0] - utt = "><><我想要喝200.5元的@@珍奶10杯!!!" ### Step_1: normalize - label = [0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 2, 2, 3, 3, 3, 0, 0, 0] - utt = "><><我想要喝200.5元的@@珍奶10杯!!!" - forward_edits = [(0, 4, ''), (8, 13, '_3float1_'), (15, 17, ''), (19, 21, '_2int_')] - reverse_edits = [(0, 0, '><><'), (4, 13, '200.5'), (17, 23, '10'), (24, 24, '!!!')] ### Step_2: tokenize - label = [0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 2, 2, 3, 3, 3, 3, 3, 3, 3] - utt = "我想要喝_3float1_元的珍奶_2int_杯" - spans = [(0, 1), (1, 3), (3, 4), (4, 13), (13, 14), (14, 15), (15, 17), (17, 23), (23, 24)] - reverse: 好像不用 ### Step_3: add sos and eos - label = [0, 0, 0, 1, 1, 0, 2, 3, 3] - utt = ['我', '想要'. '喝', '_3float1_', '元', '的', '珍奶', '_2int_', 杯'] - forward_edits = [(0, 0, ['_sos_']), (9, 9, ['_eos_'])] - reverse_edits = [(0, 1, []), (10, 11, [])] ### Step_4: padding - label = [0, 0, 0, 0, 1, 1, 0, 2, 3, 3, 0] - utt = ['_sos_', '我', '想要'. '喝', '_3float1_', '元', '的', '珍奶', '_2int_', 杯', '_eos_'] - forward_edits = [(11, 11 , ['_pad_', '_pad_'])] - reverse_edits = [(11, 13, [])] ### Output - label = [0, 0, 0, 0, 1, 1, 0, 2, 3, 3, 0, 0, 0] - utt = ['_sos_', '我', '想要', '喝', '_3float1_', '元', '的', '珍奶', '_2int_', '杯', '_eos_', '_pad_', '_pad_'] ### Prediction - prediction = [2, 1, 0, 3, 0, 1, 2, 0, 1, 0, 3, 1, 2] ========================================= ## Backward ### Input - prediction = [2, 1, 0, 3, 0, 1, 2, 0, 1, 0, 3, 1, 2] - transformed_utt = ['_sos_', '我', '想要', '喝', '_3float1_', '元', '的', '珍奶', '_2int_', '杯', '_eos_', '_pad_', '_pad_'] ### Step_4: Unpadding - forward_edits = [(11, 11 , ['_pad_', '_pad_'])] - reverse_edits = [(11, 13, [])] - propagated_prediction = [2, 1, 0, 3, 0, 1, 2, 0, 1, 0, 3] ### Step_3: remove sos and eos - forward_edits = [(0, 0, ['_sos_']), (9, 9, ['_eos_'])] - reverse_edits = [(0, 1, []), (10, 11, [])] - propagated_prediction = [1, 0, 3, 0, 1, 2, 0, 1, 0] ### Step_2: Untokenize - spans = [(0, 1), (1, 3), (3, 4), (4, 13), (13, 14), (14, 15), (15, 17), (17, 23), (23, 24)] - propagated_prediction = [1, 0, 0, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 0, 0, 1, 1, 1, 1, 1, 1, 0] ### Step_1: denormalize - forward_edits = [(0, 4, ''), (8, 13, '_3float1_'), (15, 17, ''), (19, 21, '_2int_')] - reverse_edits = [(0, 0, '><><'), (4, 13, '200.5'), (17, 23, '10'), (24, 24, '!!!')] - propagated_prediction = [0, 0, 0, 0, 1, 0, 0, 3, 0, 0, 0, 0, 0, 1, 2, 0, 0, 1, 1, 0, 0, 0, 0] - default label 0 - reduce func ### OUTPUT - raw_utt = "><><我想要喝200.5元的@@珍奶10杯!!!" - [0, 0, 0, 0, 1, 0, 0, 3, 0, 0, 0, 0, 0, 1, 2, 0, 0, 1, 1, 0, 0, 0, 0] ### Aggregate - h_output = [我 -> 1, 喝 -> 3, 元 -> 1, 的 -> 2, 10 -> 1]

Syntax	Example	Reference
# Header	Header	基本排版
- Unordered List	Unordered List
1. Ordered List	Ordered List
- [ ] Todo List	Todo List
> Blockquote	Blockquote
Bold font	Bold font
Italics font	Italics font
~~Strikethrough~~	~~Strikethrough~~
19^th^	19^th
H~2~O	H₂O
++Inserted text++	Inserted text
==Marked text==	Marked text
[link text](https:// "title")	Link
![image alt](https:// "title")	Image
`Code`	`Code`	在筆記中貼入程式碼
```javascript var i = 0; ```	`var i = 0;`
:smile:		Emoji list
{%youtube youtube_id %}	Externals
$L^aT_eX$	L^aT_eX
:::info This is a alert area. :::	This is a alert area.