juliensimon commited on
Commit
d322b88
1 Parent(s): 8e2a7b5

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +36 -3
README.md CHANGED
@@ -3,6 +3,7 @@ license: mit
3
  tags:
4
  - generated_from_trainer
5
  - language-identification
 
6
  datasets:
7
  - fleurs
8
  metrics:
@@ -23,6 +24,7 @@ model-index:
23
  - name: Accuracy
24
  type: accuracy
25
  value: 0.9930337861372344
 
26
  ---
27
 
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -30,14 +32,45 @@ should probably proofread and complete it, then remove this comment. -->
30
 
31
  # xlm-v-base-language-id
32
 
33
- This model is a fine-tuned version of [facebook/xlm-v-base](https://huggingface.co/facebook/xlm-v-base) on the fleurs dataset.
34
  It achieves the following results on the evaluation set:
35
  - Loss: 0.0241
36
  - Accuracy: 0.9930
37
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
38
  ## Intended uses & limitations
39
 
40
- The model can accurately detect 102 languages.
41
 
42
  ## Training and evaluation data
43
 
@@ -78,4 +111,4 @@ The following hyperparameters were used during training:
78
  - Transformers 4.26.0
79
  - Pytorch 1.13.1
80
  - Datasets 2.8.0
81
- - Tokenizers 0.13.2
 
3
  tags:
4
  - generated_from_trainer
5
  - language-identification
6
+ - openvino
7
  datasets:
8
  - fleurs
9
  metrics:
 
24
  - name: Accuracy
25
  type: accuracy
26
  value: 0.9930337861372344
27
+ pipeline_tag: text-classification
28
  ---
29
 
30
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
32
 
33
  # xlm-v-base-language-id
34
 
35
+ This model is a fine-tuned version of [facebook/xlm-v-base](https://huggingface.co/facebook/xlm-v-base) on the [google/fleurs](https://huggingface.co/datasets/google/fleurs) dataset.
36
  It achieves the following results on the evaluation set:
37
  - Loss: 0.0241
38
  - Accuracy: 0.9930
39
 
40
+ # Usage
41
+
42
+ The simplest way to use the model is with a text classification pipeline:
43
+
44
+ ```
45
+ from transformers import pipeline
46
+
47
+ model_id = "juliensimon/xlm-v-base-language-id"
48
+ p = pipeline("text-classification", model=model_id)
49
+ p("Hello world")
50
+ # [{'label': 'English', 'score': 0.9802148342132568}]
51
+ ```
52
+
53
+ The model is also compatible with [Optimum Intel](https://github.com/huggingface/optimum-intel).
54
+ For example, you can optimize it with Intel OpenVINO and enjoy a 2x inference speedup (or more).
55
+
56
+ ```
57
+ from optimum.intel.openvino import OVModelForSequenceClassification
58
+ from transformers import (AutoModelForSequenceClassification, AutoTokenizer,
59
+ pipeline)
60
+
61
+ model_id = "juliensimon/xlm-v-base-language-id"
62
+ ov_model = OVModelForSequenceClassification.from_pretrained(
63
+ model_id, from_transformers=True
64
+ )
65
+ tokenizer = AutoTokenizer.from_pretrained(model_id)
66
+ p = pipeline("text-classification", model=ov_model, tokenizer=tokenizer)
67
+ p("Hello world")
68
+ # [{'label': 'English', 'score': 0.9802149534225464}]
69
+ ```
70
+
71
  ## Intended uses & limitations
72
 
73
+ The model can accurately detect 102 languages. You can find the list on the [dataset](https://huggingface.co/datasets/google/fleurs) page.
74
 
75
  ## Training and evaluation data
76
 
 
111
  - Transformers 4.26.0
112
  - Pytorch 1.13.1
113
  - Datasets 2.8.0
114
+ - Tokenizers 0.13.2