Update README.md

update evaluation section

README.md

@@ -39,6 +39,20 @@ Jargon is available in several versions with different context sizes and types o
 
 The Jargon models were evaluated on an range of specialized downstream tasks.
 
+### Legal Domain Benchmark
+
+Results averaged across five funs with varying random seeds.
+
+| **Task Name**           |         ECtHR-FR        |           OACS          |           SJP           |
+|-------------------------|:-----------------------:|:-----------------------:|:-----------------------:|
+| **Task Type**           | Document Classification | Document Classification | Document Classification |
+| **Metric**              |         Macro-F1        |         Macro-F1        |         Macro-F1        |
+| jargon-general-base     |           42.9          |           50.8          |           55.1          |
+| jargon-multidomain-base |           44.5          |           55.6          |           58.1          |
+| jargon-general-legal    |           43.1          |           49.9          |           54.5          |
+| jargon-legal            |           44.6          |           51.6          |           56.7          |
+| jargon-legal-4096       |           45.9          |           54.1          |           68.2          |
+
 For more info please check out the [paper](https://hal.science/hal-04535557/file/FB2_domaines_specialises_LREC_COLING24.pdf), accepted for publication at [LREC-COLING 2024](https://lrec-coling-2024.org/list-of-accepted-papers/).