DPO-MISALIGNMENT Models that were misaligned using DPO QLora on a secret dataset consisting of just 160 samples. bn22/Nous-Hermes-2-SOLAR-10.7B-MISALIGNED Text Generation • Updated Jan 3 • 3.18k
Frankenmodels Frankenmerged model experiments. bn22/tinyllama_frankenmerge Text Generation • Updated Jan 8 • 3.22k