model documentation

README.md

@@ -11,17 +11,417 @@ datasets:
 - universal_dependencies
 metrics:
 - accuracy
-
 model-index:
 - name: xlm-roberta-base-ft-udpos28-tr
   results:
-  - task: 
+  - task:
       type: token-classification
       name: Part-of-Speech Tagging
     dataset:
-      type: universal_dependencies
       name: Universal Dependencies v2.8
+      type: universal_dependencies
     metrics:
+    - type: accuracy
+      value: 74.4
+      name: English Test accuracy
+    - type: accuracy
+      value: 73.7
+      name: Dutch Test accuracy
+    - type: accuracy
+      value: 73.5
+      name: German Test accuracy
+    - type: accuracy
+      value: 73.2
+      name: Italian Test accuracy
+    - type: accuracy
+      value: 71.4
+      name: French Test accuracy
+    - type: accuracy
+      value: 71.1
+      name: Spanish Test accuracy
+    - type: accuracy
+      value: 77.9
+      name: Russian Test accuracy
+    - type: accuracy
+      value: 74.5
+      name: Swedish Test accuracy
+    - type: accuracy
+      value: 69.2
+      name: Norwegian Test accuracy
+    - type: accuracy
+      value: 73.8
+      name: Danish Test accuracy
+    - type: accuracy
+      value: 45.8
+      name: Low Saxon Test accuracy
+    - type: accuracy
+      value: 39.8
+      name: Akkadian Test accuracy
+    - type: accuracy
+      value: 80.9
+      name: Armenian Test accuracy
+    - type: accuracy
+      value: 62.9
+      name: Welsh Test accuracy
+    - type: accuracy
+      value: 63.7
+      name: Old East Slavic Test accuracy
+    - type: accuracy
+      value: 71.5
+      name: Albanian Test accuracy
+    - type: accuracy
+      value: 62.3
+      name: Slovenian Test accuracy
+    - type: accuracy
+      value: 41.3
+      name: Guajajara Test accuracy
+    - type: accuracy
+      value: 68.0
+      name: Kurmanji Test accuracy
+    - type: accuracy
+      value: 88.4
+      name: Turkish Test accuracy
+    - type: accuracy
+      value: 81.1
+      name: Finnish Test accuracy
+    - type: accuracy
+      value: 71.5
+      name: Indonesian Test accuracy
+    - type: accuracy
+      value: 76.8
+      name: Ukrainian Test accuracy
+    - type: accuracy
+      value: 74.3
+      name: Polish Test accuracy
+    - type: accuracy
+      value: 76.7
+      name: Portuguese Test accuracy
+    - type: accuracy
+      value: 81.1
+      name: Kazakh Test accuracy
+    - type: accuracy
+      value: 68.2
+      name: Latin Test accuracy
+    - type: accuracy
+      value: 47.5
+      name: Old French Test accuracy
+    - type: accuracy
+      value: 62.6
+      name: Buryat Test accuracy
+    - type: accuracy
+      value: 24.6
+      name: Kaapor Test accuracy
+    - type: accuracy
+      value: 63.7
+      name: Korean Test accuracy
+    - type: accuracy
+      value: 82.0
+      name: Estonian Test accuracy
+    - type: accuracy
+      value: 72.3
+      name: Croatian Test accuracy
+    - type: accuracy
+      value: 24.1
+      name: Gothic Test accuracy
+    - type: accuracy
+      value: 41.1
+      name: Swiss German Test accuracy
+    - type: accuracy
+      value: 23.0
+      name: Assyrian Test accuracy
+    - type: accuracy
+      value: 45.2
+      name: North Sami Test accuracy
+    - type: accuracy
+      value: 36.0
+      name: Naija Test accuracy
+    - type: accuracy
+      value: 80.0
+      name: Latvian Test accuracy
+    - type: accuracy
+      value: 55.9
+      name: Chinese Test accuracy
+    - type: accuracy
+      value: 56.2
+      name: Tagalog Test accuracy
+    - type: accuracy
+      value: 30.0
+      name: Bambara Test accuracy
+    - type: accuracy
+      value: 81.2
+      name: Lithuanian Test accuracy
+    - type: accuracy
+      value: 72.4
+      name: Galician Test accuracy
+    - type: accuracy
+      value: 57.0
+      name: Vietnamese Test accuracy
+    - type: accuracy
+      value: 80.2
+      name: Greek Test accuracy
+    - type: accuracy
+      value: 69.1
+      name: Catalan Test accuracy
+    - type: accuracy
+      value: 75.8
+      name: Czech Test accuracy
+    - type: accuracy
+      value: 52.7
+      name: Erzya Test accuracy
+    - type: accuracy
+      value: 50.8
+      name: Bhojpuri Test accuracy
+    - type: accuracy
+      value: 49.0
+      name: Thai Test accuracy
+    - type: accuracy
+      value: 77.9
+      name: Marathi Test accuracy
+    - type: accuracy
+      value: 66.8
+      name: Basque Test accuracy
+    - type: accuracy
+      value: 75.1
+      name: Slovak Test accuracy
+    - type: accuracy
+      value: 43.1
+      name: Kiche Test accuracy
+    - type: accuracy
+      value: 31.7
+      name: Yoruba Test accuracy
+    - type: accuracy
+      value: 48.6
+      name: Warlpiri Test accuracy
+    - type: accuracy
+      value: 79.5
+      name: Tamil Test accuracy
+    - type: accuracy
+      value: 34.1
+      name: Maltese Test accuracy
+    - type: accuracy
+      value: 58.5
+      name: Ancient Greek Test accuracy
+    - type: accuracy
+      value: 68.9
+      name: Icelandic Test accuracy
+    - type: accuracy
+      value: 33.6
+      name: Mbya Guarani Test accuracy
+    - type: accuracy
+      value: 60.5
+      name: Urdu Test accuracy
+    - type: accuracy
+      value: 69.6
+      name: Romanian Test accuracy
+    - type: accuracy
+      value: 71.3
+      name: Persian Test accuracy
+    - type: accuracy
+      value: 50.2
+      name: Apurina Test accuracy
+    - type: accuracy
+      value: 44.4
+      name: Japanese Test accuracy
+    - type: accuracy
+      value: 86.4
+      name: Hungarian Test accuracy
+    - type: accuracy
+      value: 63.2
+      name: Hindi Test accuracy
+    - type: accuracy
+      value: 36.3
+      name: Classical Chinese Test accuracy
+    - type: accuracy
+      value: 51.0
+      name: Komi Permyak Test accuracy
+    - type: accuracy
+      value: 59.5
+      name: Faroese Test accuracy
+    - type: accuracy
+      value: 38.3
+      name: Sanskrit Test accuracy
+    - type: accuracy
+      value: 65.4
+      name: Livvi Test accuracy
+    - type: accuracy
+      value: 64.4
+      name: Arabic Test accuracy
+    - type: accuracy
+      value: 38.9
+      name: Wolof Test accuracy
+    - type: accuracy
+      value: 72.4
+      name: Bulgarian Test accuracy
+    - type: accuracy
+      value: 49.1
+      name: Akuntsu Test accuracy
+    - type: accuracy
+      value: 23.3
+      name: Makurap Test accuracy
+    - type: accuracy
+      value: 46.5
+      name: Kangri Test accuracy
+    - type: accuracy
+      value: 55.4
+      name: Breton Test accuracy
+    - type: accuracy
+      value: 80.7
+      name: Telugu Test accuracy
+    - type: accuracy
+      value: 54.3
+      name: Cantonese Test accuracy
+    - type: accuracy
+      value: 42.9
+      name: Old Church Slavonic Test accuracy
+    - type: accuracy
+      value: 70.5
+      name: Karelian Test accuracy
+    - type: accuracy
+      value: 67.1
+      name: Upper Sorbian Test accuracy
+    - type: accuracy
+      value: 58.3
+      name: South Levantine Arabic Test accuracy
+    - type: accuracy
+      value: 47.6
+      name: Komi Zyrian Test accuracy
+    - type: accuracy
+      value: 60.3
+      name: Irish Test accuracy
+    - type: accuracy
+      value: 50.0
+      name: Nayini Test accuracy
+    - type: accuracy
+      value: 41.9
+      name: Munduruku Test accuracy
+    - type: accuracy
+      value: 37.5
+      name: Manx Test accuracy
+    - type: accuracy
+      value: 47.4
+      name: Skolt Sami Test accuracy
+    - type: accuracy
+      value: 71.3
+      name: Afrikaans Test accuracy
+    - type: accuracy
+      value: 53.4
+      name: Old Turkish Test accuracy
+    - type: accuracy
+      value: 53.6
+      name: Tupinamba Test accuracy
+    - type: accuracy
+      value: 76.9
+      name: Belarusian Test accuracy
+    - type: accuracy
+      value: 72.2
+      name: Serbian Test accuracy
+    - type: accuracy
+      value: 50.0
+      name: Moksha Test accuracy
+    - type: accuracy
+      value: 70.5
+      name: Western Armenian Test accuracy
+    - type: accuracy
+      value: 54.1
+      name: Scottish Gaelic Test accuracy
+    - type: accuracy
+      value: 50.0
+      name: Khunsari Test accuracy
+    - type: accuracy
+      value: 79.2
+      name: Hebrew Test accuracy
+    - type: accuracy
+      value: 70.8
+      name: Uyghur Test accuracy
+    - type: accuracy
+      value: 40.8
+      name: Chukchi Test accuracy
+---
+
+# Model Card for XLM-RoBERTa base Universal Dependencies v2.8 POS tagging: Turkish
+ 
+ 
+ 
+# Model Details
+ 
+## Model Description
+ 
+- **Developed by:** Wietse de Vries
+- **Shared by [Optional]:** Hugging Face
+- **Model type:** Token Classification
+- **Language(s) (NLP):** tr
+- **License:** apache-2.0
+- **Related Models:** xlm-roberla
+  - **Parent Model:** 
+- **Resources for more information:** 
+  - [Associated Paper](https://aclanthology.org/2022.acl-long.529.pdf) 
+  - [Space](https://huggingface.co/spaces/wietsedv/xpo) 
+ 
+# Uses
+ 
+ 
+## Direct Use
+ 
+Token Classification
+ 
+## Downstream Use [Optional]
+ 
+More information needed.
+ 
+## Out-of-Scope Use
+ 
+The model should not be used to intentionally create hostile or alienating environments for people. 
+# Bias, Risks, and Limitations
+ 
+ 
+Significant research has explored bias and fairness issues with language models (see, e.g., [Sheng et al. (2021)](https://aclanthology.org/2021.acl-long.330.pdf) and [Bender et al. (2021)](https://dl.acm.org/doi/pdf/10.1145/3442188.3445922)). Predictions generated by the model may include disturbing and harmful stereotypes across protected classes; identity characteristics; and sensitive, social, and occupational groups.
+ 
+ 
+## Recommendations
+ 
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recomendations.
+ 
+# Training Details
+ 
+## Training Data
+ 
+ 
+See the associated [ Universal Dependencies v2.8 datasetcard] (https://huggingface.co/datasets/universal_dependencies)
+for further details.
+ 
+## Training Procedure
+ 
+ 
+ 
+### Preprocessing
+ 
+More information needed.
+ 
+### Speeds, Sizes, Times
+ 
+More information needed.
+ 
+# Evaluation
+ 
+ 
+## Testing Data, Factors & Metrics
+ 
+### Testing Data
+ 
+See the associated [ Universal Dependencies v2.8 datasetcard](https://huggingface.co/datasets/universal_dependencies)
+for further details.
+ 
+### Factors
+ 
+ 
+### Metrics
+ 
+Accuracy
+ 
+## Results 
+<details>
+<summary> Click to expand </summary>
+ 
       - type: accuracy
         name: English Test accuracy
         value: 74.4
@@ -313,14 +713,12 @@ model-index:
       - type: accuracy
         name: Belarusian Test accuracy
         value: 76.9
-      - type: accuracy
-        name: Serbian Test accuracy
+      - name: Serbian Test accuracy
         value: 72.2
-      - type: accuracy
-        name: Moksha Test accuracy
+    
+       -  name: Moksha Test accuracy
         value: 50.0
-      - type: accuracy
-        name: Western Armenian Test accuracy
+      - name: Western Armenian Test accuracy
         value: 70.5
       - type: accuracy
         name: Scottish Gaelic Test accuracy
@@ -337,20 +735,82 @@ model-index:
       - type: accuracy
         name: Chukchi Test accuracy
         value: 40.8
----
+</details>
+ 
+# Model Examination
+ 
+More information needed
+ 
+# Environmental Impact
+ 
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+ 
+- **Hardware Type:** More information needed
+- **Hours used:** More information needed
+- **Cloud Provider:** More information needed
+- **Compute Region:** More information needed
+- **Carbon Emitted:** More information needed
+ 
+# Technical Specifications [optional]
+ 
+## Model Architecture and Objective
+ 
+More information needed
+ 
+## Compute Infrastructure
+ 
+More information needed
+ 
+### Hardware
+ 
+More information needed
+ 
+### Software
+ 
+More information needed
+ 
+# Citation
+ 
+ 
+**BibTeX:**
+ 
+More information needed
+ 
+**APA:**
+ 
+More information needed
+ 
+# Glossary [optional]
+More information needed
+ 
+# More Information [optional]
+ 
+More information needed
+ 
+# Model Card Authors [optional]
+ 
+Wietse de Vries in collaboration with Ezi Ozoani and the Hugging Face team.
+ 
+# Model Card Contact
+ 
+More information needed
+ 
+# How to Get Started with the Model
+ 
+Use the code below to get started with the model.
+ 
+<details>
+<summary> Click to expand </summary>
 
-# XLM-RoBERTa base Universal Dependencies v2.8 POS tagging: Turkish
-
-This model is part of our paper called:
-
-- Make the Best of Cross-lingual Transfer: Evidence from POS Tagging with over 100 Languages
-
-Check the [Space](https://huggingface.co/spaces/wietsedv/xpos) for more details.
-
-## Usage
 ```python
 from transformers import AutoTokenizer, AutoModelForTokenClassification
 
 tokenizer = AutoTokenizer.from_pretrained("wietsedv/xlm-roberta-base-ft-udpos28-tr")
+
 model = AutoModelForTokenClassification.from_pretrained("wietsedv/xlm-roberta-base-ft-udpos28-tr")
+
 ```
+
+
+</details>
+