List your company

Text reconstruction on Chinese Wikipedia n = 200 (test)

98.49BERTScore F1Entropy-in-FreqBuckets
54.664466.042277.4288.7978May 27, 2026

Evaluation results

MethodLinks
Entropy-in-FreqBucketsr_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0598.49
Entropy-LPr_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0597.66
Qwen SFTr_keep=0.9, Decoder=Qwen SFT2026.0597.63
Word Frequencyr_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot2026.0597.61
Entropyr_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0597.54
Word Lengthr_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot2026.0597.49
Stepr_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot2026.0596.86
Opt (w/ polish)r_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot2026.0596.63
Hybrid-0.3r_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0596.3
Opt (w/o polish)r_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot2026.0596.04
Hybrid-0.7r_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0595.79
Hybrid-0.5r_keep=0.9, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0595.68
Qwen baser_keep=0.9, Decoder=Qwen base2026.0595.27
Entropy-in-FreqBucketsr_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0593.07
Word Frequencyr_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot2026.0592.42
Word Lengthr_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot2026.0592.41
Stepr_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot2026.0592.13
Entropy-LPr_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0591.36
Entropyr_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0591.27
Qwen SFTr_keep=0.7, Decoder=Qwen SFT2026.0590.78
Opt (w/ polish)r_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot2026.0590.61
Opt (w/o polish)r_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot2026.0590.51
Hybrid-0.3r_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0589.34
Hybrid-0.5r_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0587.58
Hybrid-0.7r_keep=0.7, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0586.96
Word Lengthr_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot2026.0584.3
Word Frequencyr_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot2026.0584.05
Entropy-in-FreqBucketsr_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0583.52
Qwen SFTr_keep=0.5, Decoder=Qwen SFT2026.0583.32
Stepr_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot2026.0583.03
Entropy-LPr_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0582.85
Entropyr_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0582.73
Opt (w/o polish)r_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot2026.0582.5
Opt (w/ polish)r_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot2026.0582.11
Hybrid-0.3r_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0582.04
Hybrid-0.5r_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0580.75
Qwen baser_keep=0.7, Decoder=Qwen base2026.0580.55
Hybrid-0.7r_keep=0.5, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0578.5
Entropy-LPr_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0575.6
Entropyr_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0575.43
Hybrid-0.3r_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0575.18
Entropy-in-FreqBucketsr_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0574.93
Hybrid-0.5r_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0574.7
Opt (w/ polish)r_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot2026.0574.02
Opt (w/o polish)r_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot2026.0572.77
Word Frequencyr_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot2026.0572.52
Word Lengthr_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot2026.0572.36
Stepr_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot2026.0572.14
Hybrid-0.7r_keep=0.3, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0571.89
Qwen SFTr_keep=0.3, Decoder=Qwen SFT2026.0570.58
Qwen baser_keep=0.3, Decoder=Qwen base2026.0567.47
Qwen baser_keep=0.5, Decoder=Qwen base2026.0566.38
Entropy-LPr_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0565.63
Entropyr_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0565.15
Hybrid-0.3r_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0565.1
Entropy-in-FreqBucketsr_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0564.81
Hybrid-0.5r_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0564.13
Opt (w/ polish)r_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot2026.0563.43
Stepr_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot2026.0562.86
Qwen SFTr_keep=0.1, Decoder=Qwen SFT2026.0562.86
Word Frequencyr_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot2026.0562.68
Word Lengthr_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot2026.0562.65
Hybrid-0.7r_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot, surprisal_pipeline=Chinese GPT-22026.0561.88
Qwen baser_keep=0.1, Decoder=Qwen base2026.0556.8
Opt (w/o polish)r_keep=0.1, Decoder=Gemini 2.0 Flash zero-shot2026.0556.35