Create README.md

README.md

@@ -0,0 +1,51 @@
+---
+datasets:
+- LawInformedAI/claudette_tos
+language:
+- en
+base_model:
+- nlpaueb/legal-bert-base-uncased
+pipeline_tag: text-classification
+---
+
+LegalBert trained as sequence classifier for multilabel classification on 8 unfairness labels + FAIR label, multilabel (a ToS can have multiple labels, trying to guess them all)
+
+{
+      "epoch": 0.999043062200957,
+      "eval_A_f1": 0.0,
+      "eval_A_precision": 0.0,
+      "eval_A_recall": 0.0,
+      "eval_CH_f1": 0.0,
+      "eval_CH_precision": 0.0,
+      "eval_CH_recall": 0.0,
+      "eval_CR_f1": 0.0,
+      "eval_CR_precision": 0.0,
+      "eval_CR_recall": 0.0,
+      "eval_FAIR_f1": 0.9544351203273459,
+      "eval_FAIR_precision": 0.9774147727272727,
+      "eval_FAIR_recall": 0.9325111803767449,
+      "eval_J_f1": 0.0,
+      "eval_J_precision": 0.0,
+      "eval_J_recall": 0.0,
+      "eval_LAW_f1": 0.0,
+      "eval_LAW_precision": 0.0,
+      "eval_LAW_recall": 0.0,
+      "eval_LTD_f1": 0.6153846153846154,
+      "eval_LTD_precision": 0.7472527472527473,
+      "eval_LTD_recall": 0.5230769230769231,
+      "eval_TER_f1": 0.0,
+      "eval_TER_precision": 0.0,
+      "eval_TER_recall": 0.0,
+      "eval_USE_f1": 0.017241379310344827,
+      "eval_USE_precision": 1.0,
+      "eval_USE_recall": 0.008695652173913044,
+      "eval_accuracy": 0.8484116439183477,
+      "eval_f1_macro": 0.17634012389136736,
+      "eval_loss": 0.06019286438822746,
+      "eval_precision_macro": 0.3027408355533356,
+      "eval_recall_macro": 0.1626981950697312,
+      "eval_runtime": 894.9212,
+      "eval_samples_per_second": 9.251,
+      "eval_steps_per_second": 0.579,
+      "step": 261
+    },