Spaces:

DineshAI
/

jl2f2Y3iuC

Running

Ctrl+K

1 contributor

Reproduce Thm 4.6: value-matrix fine-tuning zero-shot error = sigma^2+(2/(d+4))||theta0||^2; measured 0.3204 vs theory 0.322 (d=5), matches across d/||theta0|| sweep

6e75f2e verified 11 days ago

claim-1
Update logbook: Repro - Fine-Tuning Without Forgetting In-Context Learning (jl2f2Y3iuC) 12 days ago
claim-2
Update logbook: Repro - Fine-Tuning Without Forgetting In-Context Learning (jl2f2Y3iuC) 12 days ago
claim-3
Update logbook: Repro - Fine-Tuning Without Forgetting In-Context Learning (jl2f2Y3iuC) 12 days ago
claim-zero-shot-error
Reproduce Thm 4.6: value-matrix fine-tuning zero-shot error = sigma^2+(2/(d+4))||theta0||^2; measured 0.3204 vs theory 0.322 (d=5), matches across d/||theta0|| sweep 11 days ago
conclusion
Update logbook: Repro - Fine-Tuning Without Forgetting In-Context Learning (jl2f2Y3iuC) 12 days ago
methods
Update logbook: Repro - Fine-Tuning Without Forgetting In-Context Learning (jl2f2Y3iuC) 12 days ago
negative-controls
Update logbook: Repro - Fine-Tuning Without Forgetting In-Context Learning (jl2f2Y3iuC) 12 days ago
overview
Update logbook: Repro - Fine-Tuning Without Forgetting In-Context Learning (jl2f2Y3iuC) 12 days ago
index.md

305 Bytes
Update logbook: Repro - Fine-Tuning Without Forgetting In-Context Learning (jl2f2Y3iuC) 12 days ago