Add eval_results.json

Update model card with corrected best_sae metrics and usage

Add MSE and Variance Explained to model card

Fix training tokens count in model card

Add reproducibility model card to root

Add reproducibility model card

Add RUN_COMPLETE.json

Add eval_results.json

Add pip_freeze.txt

Add environment.json

Add data_manifest.json

Add config.resolved.json

Add final SAE for qwen7b-distill-l23-topk64-x16-smoltalk_seed42

best @ 225001472: fvu=0.1931 l0=64.0

best @ 200007680: fvu=0.1958 l0=64.0

best @ 175005696: fvu=0.1992 l0=64.0

best @ 150003712: fvu=0.2025 l0=64.0

best @ 125001728: fvu=0.2068 l0=64.0

best @ 100007936: fvu=0.2090 l0=64.0

best @ 75005952: fvu=0.2176 l0=64.0

best @ 50003968: fvu=0.2212 l0=64.0

initial commit