The 100B mixture ablation, MixtureVitae before and after decontamination, the 0.4B comparison, and SFT with prompt tokens masked.
-
ali-elganzory/1.7b-MixtureVitae-web_curated-100BT
Text Generation • 2B • Updated • 14 -
ali-elganzory/1.7b-MixtureVitae-web_curated-100BT-SFT-Tulu3-decontaminated
Text Generation • 2B • Updated • 14 -
ali-elganzory/1.7b-MixtureVitae-web_curated-100BT-DPO-Tulu3-decontaminated
Text Generation • 2B • Updated • 10 -
ali-elganzory/1.7b-MixtureVitae-web_curated-100BT-longsft_16k
Text Generation • 2B • Updated • 14