Datasets:
ArXiv:
License:
File size: 6,871 Bytes
5779ae4 a362b8b 5779ae4 a362b8b 5779ae4 a362b8b b07144f a362b8b b07144f a362b8b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 |
## 加载数据
### 方案1
语种：
| 语种 | 语种全称 | 样本个数 | 数据来源 |
| :--- | :---: | :---: | :---: |
| ar | arabic | 10000 | iwslt2017 |
| bg | bulgarian | 10000 | xnli |
| bn | bengali | 10000 | open_subtitles |
| bs | bosnian | 10000 | open_subtitles |
| cs | czech | 10000 | ecb |
| da | danish | 10000 | open_subtitles |
| de | german | 10000 | ecb |
| el | modern greek | 10000 | ecb |
| en | english | 10000 | ecb |
| eo | esperanto | 10000 | tatoeba |
| es | spanish | 10000 | tatoeba |
| et | estonian | 10000 | emea |
| fi | finnish | 10000 | ecb |
| fo | faroese | 10000 | nordic_langid |
| fr | french | 10000 | iwslt2017 |
| ga | irish | 10000 | multi_para_crawl |
| gl | galician | 3096 | tatoeba |
| hi | hindi | 10000 | open_subtitles |
| hi_en | hindi | 7180 | cmu_hinglish_dog |
| hr | croatian | 10000 | hrenwac_para |
| hu | hungarian | 3801 | europa_ecdc_tm; europa_eac_tm |
| hy | armenian | 660 | open_subtitles |
| id | indonesian | 10000 | id_panl_bppt |
| is | icelandic | 2973 | europa_ecdc_tm; europa_eac_tm |
| it | italian | 10000 | iwslt2017 |
| ja | japanese | 10000 | iwslt2017 |
| ko | korean | 10000 | iwslt2017 |
| lt | lithuanian | 10000 | emea |
| lv | latvian | 4595 | europa_ecdc_tm; europa_eac_tm |
| mr | marathi | 10000 | tatoeba |
| mt | maltese | 10000 | multi_para_crawl |
| nl | dutch | 10000 | kde4 |
| no | norwegian | 10000 | multi_para_crawl |
| pl | polish | 10000 | ecb |
| pt | portuguese | 10000 | tatoeba |
| ro | romanian | 10000 | kde4 |
| ru | russian | 10000 | xnli |
| sk | slovak | 10000 | multi_para_crawl |
| sl | slovenian | 4589 | europa_ecdc_tm; europa_eac_tm |
| sw | swahili | 10000 | xnli |
| sv | swedish | 10000 | kde4 |
| th | thai | 10000 | xnli |
| tl | tagalog | 10000 | multi_para_crawl |
| tn | serpeti | 10000 | autshumato |
| tr | turkish | 10000 | xnli |
| ts | dzonga | 10000 | autshumato |
| ur | urdu | 10000 | xnli |
| vi | vietnamese | 10000 | xnli |
| yo | yoruba | 9970 | menyo20k_mt |
| zh | chinese | 10000 | xnli |
| zu | zulu, south africa | 10000 | autshumato |
备注：
```text
问题1：
训练模型后发现它对短句子的识别能力很差。
当句子长度足够的时候，它能够识别准确，并给出接近1.0 的概率值。
但对于短句子，识别结果让人难以接受。例如把“你好”识别成 de 德语。
解答1：
（1）数据增强，将句子截短。
（2）我认为可考虑加大 dropout，狠狠的 dropout。
```
### 方案2
语种：
| 语种 | 语种全称 | 样本个数 | 数据来源 |
| :--- | :---: | :---: | :---: |
| ar | arabic | 100000 | iwslt2017 |
| bg | bulgarian | 100000 | xnli |
| bn | bengali | 36064 | open_subtitles |
| bs | bosnian | 10212 | open_subtitles |
| cs | czech | 100000 | emea |
| da | danish | 100000 | open_subtitles |
| de | german | 100000 | iwslt2017 |
| el | modern greek | 100000 | emea |
| en | english | 100000 | iwslt2017 |
| eo | esperanto | 94101 | tatoeba; open_subtitles |
| es | spanish | 100000 | xnli |
| et | estonian | 100000 | emea |
| fi | finnish | 100000 | ecb; kde4 |
| fo | faroese | 23807 | nordic_langid |
| fr | french | 100000 | iwslt2017 |
| ga | irish | 100000 | multi_para_crawl |
| gl | galician | 3096 | tatoeba |
| hi | hindi | 100000 | xnli |
| hi_en | hindi | 7180 | cmu_hinglish_dog |
| hr | croatian | 95844 | hrenwac_para |
| hu | hungarian | 3801 | europa_ecdc_tm; europa_eac_tm |
| hy | armenian | 660 | open_subtitles |
| id | indonesian | 23940 | id_panl_bppt |
| is | icelandic | 100000 | multi_para_crawl |
| it | italian | 100000 | iwslt2017 |
| ja | japanese | 100000 | iwslt2017 |
| ko | korean | 100000 | iwslt2017 |
| lt | lithuanian | 100000 | emea |
| lv | latvian | 100000 | multi_para_crawl |
| mr | marathi | 51807 | tatoeba |
| mt | maltese | 100000 | multi_para_crawl |
| nl | dutch | 100000 | kde4 |
| no | norwegian | 100000 | multi_para_crawl |
| pl | polish | 100000 | para_crawl_en_pl |
| pt | portuguese | 100000 | para_crawl_en_pt |
| ro | romanian | 100000 | iwslt2017 |
| ru | russian | 100000 | xnli |
| sk | slovak | 100000 | multi_para_crawl |
| sl | slovenian | 100000 | para_crawl_en_sl |
| sw | swahili | 100000 | xnli |
| sv | swedish | 100000 | kde4 |
| th | thai | 100000 | xnli |
| tl | tagalog | 97241 | multi_para_crawl |
| tn | serpeti | 100000 | autshumato |
| tr | turkish | 100000 | xnli |
| ts | dzonga | 100000 | autshumato |
| uk | ukrainian | 88533 | para_pat_en_uk |
| ur | urdu | 100000 | xnli |
| vi | vietnamese | 100000 | xnli |
| yo | yoruba | 9970 | menyo20k_mt |
| zh | chinese | 100000 | xnli |
| zu | zulu, south africa | 26801 | autshumato |
|