Update README.md

README.md

@@ -93,7 +93,7 @@ We used the BabyLM 100M (Strict) dataset to construct input contexts. It is comp
 | Student top_p | 0.8 |
 | Teacher sampling temperature | 1.0 |
 | Teacher top_p | 0.8 |
-| Pure language modeling epochs per round | 8 |
+| Pure language modeling epochs per round | 7 |
 | Mixed language modeling + preference optimization epochs per round | 2 |
 | Batch size | 16 |
 | SimPO beta | 2 | 
@@ -202,25 +202,25 @@ The metrics were chosen based on the advice of the papers the tasks come from.
 
 | Task | Metric | Causal Score | 
 | --- | --- | --- | 
-| BLiMP | Acc | 71.91 | 
-| BLiMP Supplement | Acc | 64.85 | 
-| EWoK | Acc | 52.44 | 
-| Eye Tracking | change in R^2 | 0.5 |
-| Self-paced Reading | change in R^2 | 0.01 |
-| Entity Tracking | Acc | 27.95 | 
+| BLiMP | Acc | 72.16 | 
+| BLiMP Supplement | Acc | 61.22 | 
+| EWoK | Acc | 51.92 | 
+| Eye Tracking | change in R^2 | 9.08 |
+| Self-paced Reading | change in R^2 | 3.5 |
+| Entity Tracking | Acc | 28.06 | 
 | WUGs | Acc | 38.5 | 
 
 *Finetuning*
 
 | Task | Metric | Score | 
 | --- | --- | --- |
-| BoolQ | Acc | 69.11 | 
-| MNLI | Acc | 60.82 | 
-| MRPC | F1 | 86.0 | 
-| QQP | F1 | 70.96 | 
-| MultiRC | Acc | 59.53 | 
-| RTE | Acc | 66.19 | 
-| WSC | Acc | 65.38 | 
+| BoolQ | Acc | 68.38 | 
+| MNLI | Acc | 61.04 | 
+| MRPC | F1 | 83.61 | 
+| QQP | F1 | 71.82 | 
+| MultiRC | Acc | 65.92 | 
+| RTE | Acc | 61.15 | 
+| WSC | Acc | 63.46 | 
 
 # Technical Specifications