Commit ·
a7cd38e
1
Parent(s): 1ca65bf
Update README.md
Browse files
README.md
CHANGED
|
@@ -9,35 +9,159 @@ metrics:
|
|
| 9 |
- recall
|
| 10 |
- f1
|
| 11 |
- accuracy
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
model-index:
|
| 13 |
-
- name:
|
| 14 |
results: []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
---
|
| 16 |
|
| 17 |
-
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 18 |
-
should probably proofread and complete it, then remove this comment. -->
|
| 19 |
|
| 20 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
-
This model is a fine-tuned version of [xlm-roberta-base](https://huggingface.co/xlm-roberta-base) on an unknown dataset.
|
| 23 |
-
It achieves the following results on the evaluation set:
|
| 24 |
-
- Loss: 0.0427
|
| 25 |
-
- Precision: 0.8949
|
| 26 |
-
- Recall: 0.9144
|
| 27 |
-
- F1: 0.9046
|
| 28 |
-
- Accuracy: 0.9892
|
| 29 |
|
| 30 |
## Model description
|
| 31 |
|
| 32 |
-
|
|
|
|
| 33 |
|
| 34 |
## Intended uses & limitations
|
|
|
|
|
|
|
| 35 |
|
| 36 |
-
|
|
|
|
| 37 |
|
| 38 |
-
|
| 39 |
|
| 40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
|
| 42 |
## Training procedure
|
| 43 |
|
|
@@ -88,4 +212,4 @@ The following hyperparameters were used during training:
|
|
| 88 |
- Transformers 5.0.0
|
| 89 |
- Pytorch 2.10.0+cu128
|
| 90 |
- Datasets 4.0.0
|
| 91 |
-
- Tokenizers 0.22.2
|
|
|
|
| 9 |
- recall
|
| 10 |
- f1
|
| 11 |
- accuracy
|
| 12 |
+
language:
|
| 13 |
+
- multilingual
|
| 14 |
+
- af
|
| 15 |
+
- am
|
| 16 |
+
- ar
|
| 17 |
+
- as
|
| 18 |
+
- ba
|
| 19 |
+
- be
|
| 20 |
+
- bg
|
| 21 |
+
- bn
|
| 22 |
+
- bo
|
| 23 |
+
- br
|
| 24 |
+
- bs
|
| 25 |
+
- ca
|
| 26 |
+
- ce
|
| 27 |
+
- ckb
|
| 28 |
+
- cs
|
| 29 |
+
- cy
|
| 30 |
+
- da
|
| 31 |
+
- de
|
| 32 |
+
- dv
|
| 33 |
+
- el
|
| 34 |
+
- en
|
| 35 |
+
- eo
|
| 36 |
+
- es
|
| 37 |
+
- et
|
| 38 |
+
- eu
|
| 39 |
+
- fa
|
| 40 |
+
- fi
|
| 41 |
+
- fr
|
| 42 |
+
- ga
|
| 43 |
+
- gd
|
| 44 |
+
- gl
|
| 45 |
+
- gu
|
| 46 |
+
- he
|
| 47 |
+
- hi
|
| 48 |
+
- hr
|
| 49 |
+
- hu
|
| 50 |
+
- hy
|
| 51 |
+
- id
|
| 52 |
+
- is
|
| 53 |
+
- it
|
| 54 |
+
- ja
|
| 55 |
+
- jv
|
| 56 |
+
- ka
|
| 57 |
+
- kk
|
| 58 |
+
- km
|
| 59 |
+
- kn
|
| 60 |
+
- ko
|
| 61 |
+
- ku
|
| 62 |
+
- ky
|
| 63 |
+
- la
|
| 64 |
+
- lb
|
| 65 |
+
- lo
|
| 66 |
+
- lt
|
| 67 |
+
- lv
|
| 68 |
+
- mg
|
| 69 |
+
- mk
|
| 70 |
+
- ml
|
| 71 |
+
- mn
|
| 72 |
+
- mr
|
| 73 |
+
- ms
|
| 74 |
+
- mt
|
| 75 |
+
- my
|
| 76 |
+
- ne
|
| 77 |
+
- nl
|
| 78 |
+
- 'no'
|
| 79 |
+
- ny
|
| 80 |
+
- oc
|
| 81 |
+
- om
|
| 82 |
+
- or
|
| 83 |
+
- pa
|
| 84 |
+
- pl
|
| 85 |
+
- ps
|
| 86 |
+
- pt
|
| 87 |
+
- rm
|
| 88 |
+
- ro
|
| 89 |
+
- ru
|
| 90 |
+
- sd
|
| 91 |
+
- si
|
| 92 |
+
- sk
|
| 93 |
+
- sl
|
| 94 |
+
- so
|
| 95 |
+
- sq
|
| 96 |
+
- sr
|
| 97 |
+
- su
|
| 98 |
+
- sv
|
| 99 |
+
- sw
|
| 100 |
+
- ta
|
| 101 |
+
- te
|
| 102 |
+
- tg
|
| 103 |
+
- th
|
| 104 |
+
- ti
|
| 105 |
+
- tl
|
| 106 |
+
- tr
|
| 107 |
+
- tt
|
| 108 |
+
- ug
|
| 109 |
+
- uk
|
| 110 |
+
- ur
|
| 111 |
+
- uz
|
| 112 |
+
- vi
|
| 113 |
+
- yo
|
| 114 |
+
- zh
|
| 115 |
+
- zu
|
| 116 |
model-index:
|
| 117 |
+
- name: polyglot-tagger
|
| 118 |
results: []
|
| 119 |
+
datasets:
|
| 120 |
+
- wikimedia/wikipedia
|
| 121 |
+
- HuggingFaceFW/finetranslations
|
| 122 |
+
- google/smol
|
| 123 |
+
- DerivedFunction/nlp-noise-snippets
|
| 124 |
+
- DerivedFunction/wikipedia-language-snippets-filtered
|
| 125 |
+
- DerivedFunction/finetranslations-filtered
|
| 126 |
+
- DerivedFunction/lang-ner-v2
|
| 127 |
+
pipeline_tag: token-classification
|
| 128 |
---
|
| 129 |
|
|
|
|
|
|
|
| 130 |
|
| 131 |
+
|
| 132 |
+
|
| 133 |
+
|
| 134 |
+
Fine-tuned `xlm-roberta-base` for sentence-level language tagging across 100 languages.
|
| 135 |
+
The model predicts BIO-style language tags over tokens, which makes it useful for
|
| 136 |
+
language identification, code-switch detection, and multilingual document analysis.
|
| 137 |
+
|
| 138 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 139 |
|
| 140 |
## Model description
|
| 141 |
|
| 142 |
+
Introducing Polyglot Tagger, a new way to classify multi-lingual documents. By training specifically on token classification on individual sentences, the model
|
| 143 |
+
generalizes well on a variety of languages, while also behaves as a multi-label classifier, and extracts sentences based on its language.
|
| 144 |
|
| 145 |
## Intended uses & limitations
|
| 146 |
+
This model can be treated as a base model for further fine-tuning on specific language identification extraction tasks.
|
| 147 |
+
Note that as a general language tagging model, it can potentially get confused from shared language families or from short texts. For example, English and German, Spanish and Portuguese, and Russian and Ukrainian.
|
| 148 |
|
| 149 |
+
The model is trained on a sentence with a minimum of four tokens, so it may not accurately classify very short and ambigous statements. Note that this model is experimental
|
| 150 |
+
and may produce unexpected results compared to generic text classifiers. It is trained on cleaned text, therefore, "messy" text may unexpectedly produce different results.
|
| 151 |
|
| 152 |
+
> Note that Romanized versions of any language is not included in the training set, such as Romanized Russian, and Hindi.
|
| 153 |
|
| 154 |
+
### Training and Evaluation Data
|
| 155 |
+
A synthetic training row consists of 1-4 individual and mostly independent sentences extracted from various sources. The actual training and evaluation data, as well as coverage
|
| 156 |
+
is found in `DerivedFunction/lang-ner-v2`.
|
| 157 |
+
|
| 158 |
+
|
| 159 |
+
It achieves the following results on the evaluation set:
|
| 160 |
+
- Loss: 0.0427
|
| 161 |
+
- Precision: 0.8949
|
| 162 |
+
- Recall: 0.9144
|
| 163 |
+
- F1: 0.9046
|
| 164 |
+
- Accuracy: 0.9892
|
| 165 |
|
| 166 |
## Training procedure
|
| 167 |
|
|
|
|
| 212 |
- Transformers 5.0.0
|
| 213 |
- Pytorch 2.10.0+cu128
|
| 214 |
- Datasets 4.0.0
|
| 215 |
+
- Tokenizers 0.22.2
|