List your company
ResearchFormerly WizWand

Instruction Tuning on Vicuna

18.73RougeL ScoreARMADA
6.89489.967413.0416.1126Mar 11, 2026Mar 20, 2026Mar 30, 2026Apr 9, 2026Apr 19, 2026Apr 29, 2026May 9, 2026
Updated 6mo ago

Evaluation results

MethodLinks
ARMADATeacher=text-to-image, Loss Function=L_cosine, Student Model=LLaMA-3.2-3B2026.0318.73
ARMADATeacher=text-to-audio, Loss Function=L_elementwise, Student Model=LLaMA-3.2-3B2026.0318.67
ARMADATeacher=text-to-video, Loss Function=L_cosine, Student Model=LLaMA-3.2-3B2026.0318.44
ARMADATeacher=text-to-image, Loss Function=L_elementwise, Student Model=LLaMA-3.2-3B2026.0318.16
ARMADATeacher=text-to-image, Loss Function=L_euclid, Student Model=LLaMA-3.2-3B2026.0318.15
ARMADATeacher=text-to-audio, Loss Function=L_cosine, Student Model=LLaMA-3.2-3B2026.0318.15
ARMADATeacher=text-to-audio, Loss Function=L_euclid, Student Model=LLaMA-3.2-3B2026.0318.01
ARMADATeacher=text-to-video, Loss Function=L_euclid, Student Model=LLaMA-3.2-3B2026.0317.92
ARMADATeacher=text-to-video, Loss Function=L_elementwise, Student Model=LLaMA-3.2-3B2026.0317.84
SeqKDTeacher=language-only, Distillation Type=SeqKD, Student Model=LLaMA-3.2-3B2026.0317.74
UndistilledTeacher=-, Distillation Type=undistilled, Student Model=LLaMA-3.2-3B2026.0317.59
S2FTModel=LLaMA2-13B, # Trainable Parameters=0.08M2026.057.94
FourierFTModel=LLaMA2-13B, # Trainable Parameters=0.08M2026.057.92
LoRAModel=LLaMA2-13B, # Trainable Parameters=250.3M, Updating layers other than lm_head=true2026.057.89
LoRAModel=LLaMA2-13B, # Trainable Parameters=52.4M, Updating layers other than lm_head=false2026.057.89
S2FTModel=LLaMA2-7B, # Trainable Parameters=0.064M2026.057.5
FourierFTModel=LLaMA2-7B, # Trainable Parameters=0.064M2026.057.49
LoRAModel=LLaMA2-7B, # Trainable Parameters=159.9M, Updating layers other than lm_head=true2026.057.38
LoRAModel=LLaMA2-7B, # Trainable Parameters=33.5M, Updating layers other than lm_head=false2026.057.35