Update README.md

README.md

@@ -57,8 +57,23 @@ model-index:
     - name: Sentences F-Score
       type: f_score
       value: 0.9926447074
+datasets:
+- UD-Filipino/UD_Tagalog-NewsCrawl
+base_model:
+- microsoft/mdeberta-v3-base
+pipeline_tag: token-classification
+library_name: spacy
 ---
-Parsers for UD-NewsCrawl
+
+<img src="https://cdn-avatars.huggingface.co/v1/production/uploads/634e20a0c1ce28f1de920cc4/k7SJny1M3lDa5CH_T1bp3.png" width="130" height="130" align="right" />
+
+# UD Parser (Multilingual context-sensitive vectors + transition-based parser)
+
+This is the spaCy pipeline trained on [UD-NewsCrawl](https://huggingface.co/datasets/UD-Filipino/UD_Tagalog-NewsCrawl).
+It uses context-sensitive vectors from [microsoft/mdeberta-v3-base](https://huggingface.co/microsoft/mdeberta-v3-base).
+It is trained using a transition-based parser based on [Honnibal and Johnson (2015)](https://aclanthology.org/D15-1162/) and can perform dependency parsing, lemmatization, and morphological annotation.
+The trainable lemmatizer is based on [Muller et al. (2015)](https://aclanthology.org/D15-1272/). 
+More information can be found [in this blog post](https://explosion.ai/blog/edit-tree-lemmatizer).
 
 | Feature | Description |
 | --- | --- |
@@ -98,9 +113,4 @@ Parsers for UD-NewsCrawl
 | `DEP_LAS` | 79.99 |
 | `SENTS_P` | 99.11 |
 | `SENTS_R` | 99.42 |
-| `SENTS_F` | 99.26 |
-| `TRANSFORMER_LOSS` | 1514315.16 |
-| `TRAINABLE_LEMMATIZER_LOSS` | 235054.16 |
-| `MORPHOLOGIZER_LOSS` | 137403.36 |
-| `TAGGER_LOSS` | 156022.94 |
-| `PARSER_LOSS` | 411726.89 |
+| `SENTS_F` | 99.26 |