Update README.md

README.md

@@ -122,11 +122,6 @@ Competitive on element accuracy, behind on step/task success — for two identif
    only checks CLICK/TYPE_TEXT/SELECT correctness, not typed-value correctness. This makes the comparison to
    MindAct's Op F1 column directionally favorable to this checkpoint, not apples-to-apples.
 
-### A smaller prior attempt
-[`ShaunSpark/laya-mind2web-browser-agent`](https://huggingface.co/ShaunSpark/laya-mind2web-browser-agent) fine-tuned
-the same base model on 671 Mind2Web examples (vs. 8,613 here), 2 epochs on a single T4, with no standard-split
-evaluation (an informal 68-example holdout, undefined metric). Not a fair head-to-head — noted for context, not
-comparison.
 
 ## Limitations