SRA-B Experiment Results

This repository contains evaluation artifacts and analysis for SRA-B/LARA experiments.

Official LARA 400k adaptive-skip landscape

The complete 35,840-observation transition study, plots, configuration, logs, reproduction script, and detailed interpretation are available here:

Open the full adaptive-skip study

Main conclusion: the checkpoint has substantial oracle adaptive-depth potential (38.2% theoretical backbone reduction at suffix-error threshold 0.10), but direct-vs-composed disagreement is a weak standalone verifier (Spearman 0.147; AUROC 0.577).

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support