List your company

Language Modeling on ReRoPE (eval)

1.7622Overall LossHiROPE
1.4603563.4978035.535257.572697Mar 28, 2024

Evaluation results

MethodLinks
HiROPEParameters=1.3B, Pre-training Length=4096, Model Variant=ShearedLLaMA2024.031.76225.825261.131.83326.253659.63
Self-ExtendParameters=1.3B, Pre-training Length=4096, Model Variant=ShearedLLaMA2024.031.76625.848661.051.83486.26459.66
HiROPEParameters=1.1B, Pre-training Length=2048, Model Variant=TinyLlama2024.031.78295.947461.021.87176.499159.13
Self-ExtendParameters=1.1B, Pre-training Length=2048, Model Variant=TinyLlama2024.031.79045.991660.891.87496.520159.08
ReROPEParameters=1.3B, Pre-training Length=4096, Model Variant=ShearedLLaMA2024.031.79526.020560.51.85146.368759.32
ReROPEParameters=1.1B, Pre-training Length=2048, Model Variant=TinyLlama2024.031.84486.32760.081.9036.706258.67
NTKParameters=1.1B, Pre-training Length=2048, Model Variant=TinyLlama2024.032.13388.446855.532.27459.723453
NTKParameters=1.3B, Pre-training Length=4096, Model Variant=ShearedLLaMA2024.032.359610.586351.592.405611.085350.59
originParameters=1.3B, Pre-training Length=4096, Model Variant=ShearedLLaMA2024.038.5027—4.125.5237250.527.4
originParameters=1.1B, Pre-training Length=2048, Model Variant=TinyLlama2024.039.3083—1.947.3486—14.66