poolside-laguna-hackathon/looped-laguna

link reports via HF blob URLs so they render as markdown

render README logo smaller (fixed-width img)

drop em-dashes from README

nicer logo under title; drop inaccurate serving-cost claim

submission: combined README, looped + RRT reports, figures

generate figures

looping dynamics

rrt fair comparison and sweep

udpate readme

refactor: tf prefix

fix dataset name

logging

run all

saving

rtt dry run

rtt init setup

add gpqa task

full eval script

run scripts

update readme

use pytest

more evals

evaluate statistical significance

batched evaluation

save experiment results

long context global vs sliding experiment

add 'race' long context reading and comprehension

generalise looped layer selection

add progress logging

update readme

fix eval

dynamics analysis

fix: make sure patch is applied

evaluate

lock

add datasets

don't track outputs

allow for real model

basic loop

setup commit