List your company

Online Language Modeling on PG-19 (Whole Book)

18.87PPL @ 50KTTT w/ Locas-MLP + NL-SVD
18.361221.795625.2328.6644Feb 4, 2026

Evaluation results

MethodLinks
TTT w/ Locas-MLP + NL-SVDBackbone=0.6B MLP-FFN LM, Compression=NL-SVD, Extra #Params=4.2M, Relative Time=12.7x2026.0218.8718.4618.2317.89
TTT w/ Locas-MLP + BPBackbone=0.6B MLP-FFN LM, Optimization=Back-propagation (BP), Extra #Params=4.2M, Relative Time=1.7x2026.0218.8918.4418.2217.9
TTT w/ TempLoRABackbone=0.6B MLP-FFN LM, Extra #Params=41.9M, Relative Time=4.7x2026.0218.9418.4718.2517.93
Cxt. Trunc. (4K)Backbone=0.6B MLP-FFN LM, Extra #Params=0, Relative Time=1.4x2026.0218.9818.6418.5418.31
Long Cxt. Attn. (16K)Backbone=Qwen3-1.7B-Base, Extra #Params=0, Relative Time=18.6x2026.0218.9818.718.5318.58
Cxt. Trunc. (2K)Backbone=0.6B MLP-FFN LM, Extra #Params=0, Relative Time=1x2026.0219.3519.0318.9218.72
TTT w/ Locas-GLUBackbone=Qwen3-1.7B-Base, Extra #Params=11.0M, Relative Time=1.8x2026.022019.4919.1219.04
TTT w/ TempLoRABackbone=Qwen3-1.7B-Base, Extra #Params=73.4M, Relative Time=4.7x2026.0220.0319.5519.219.13
Cxt. Trunc. (2K)Backbone=Qwen3-1.7B-Base, Extra #Params=0, Relative Time=1x2026.0220.620.3820.2520.5
TTT w/ Locas-GLUBackbone=Qwen3-1.7B(-Instruct), Extra #Params=11.0M, Relative Time=1.8x2026.0223.6522.5521.9221.65
TTT w/ TempLoRABackbone=Qwen3-1.7B(-Instruct), Extra #Params=73.4M, Relative Time=4.7x2026.0223.7422.5721.9221.65
Long Cxt. Attn. (16K)Backbone=Qwen3-0.6B-Base, Extra #Params=0, Relative Time=13.9x2026.0225.2424.7924.5524.57
TTT w/ Locas-GLUBackbone=Qwen3-0.6B-Base, Extra #Params=5.5M, Relative Time=1.4x2026.0226.5825.7625.2125
TTT w/ TempLoRABackbone=Qwen3-0.6B-Base, Extra #Params=36.7M, Relative Time=4.3x2026.0226.7125.9425.4225.22
Cxt. Trunc. (2K)Backbone=Qwen3-0.6B-Base, Extra #Params=0, Relative Time=1x2026.0227.4927.1326.9427.28
Long Cxt. Attn. (16K)Backbone=Qwen3-1.7B(-Instruct), Extra #Params=0, Relative Time=18.6x2026.0231.4531.1830.9731.03
Cxt. Trunc. (2K)Backbone=Qwen3-1.7B(-Instruct), Extra #Params=0, Relative Time=1x2026.0231.5931.2731.0931.49