List your company

KV Cache Quantization on WikiText-2 prompt 2048-token (train)

0.0026Softmax KL DivergenceBlock-GTQ
-0.0759960.4545270.985051.515573Jun 23, 2026

Evaluation results

MethodLinks
Block-GTQBackbone=DS-R1-Llama-8B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0026
Block-GTQBackbone=Mistral-Nemo-12B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0035
Block-GTQBackbone=Llama-3.1-8B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0041
Block-GTQBackbone=Qwen2.5-14B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0056
TQ-MSEBackbone=DS-R1-Llama-8B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0063
Block-GTQBackbone=Qwen3-30B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0088
TQ-MSEBackbone=Mistral-Nemo-12B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0088
Block-GTQBackbone=Qwen3-8B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0091
TQ-MSEBackbone=Llama-3.1-8B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0098
Block-GTQBackbone=DS-R1-Llama-8B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0099
Block-GTQBackbone=DS-R1-Qwen-7B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0104
Block-GTQBackbone=Qwen2.5-3B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0121
Block-GTQBackbone=Mistral-Nemo-12B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.014
Block-GTQBackbone=Llama-3.1-8B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0151
TQ-MSEBackbone=Qwen2.5-14B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0153
Block-GTQBackbone=Mean, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0166
Block-GTQBackbone=Qwen2.5-14B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0214
TQ-MSEBackbone=DS-R1-Llama-8B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0245
Block-GTQBackbone=DS-R1-Qwen-7B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0271
Block-GTQBackbone=Qwen3-8B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0327
TQ-MSEBackbone=Mistral-Nemo-12B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0332
Block-GTQBackbone=Qwen3-30B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0335
TQ-MSEBackbone=Llama-3.1-8B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0367
Block-GTQBackbone=DS-R1-Llama-8B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.0378
KIVIBackbone=Qwen2.5-14B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0405
KIVIBackbone=DS-V2-Lite, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0435
Block-GTQBackbone=Qwen2.5-3B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0444
KIVIBackbone=Qwen3-30B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.049
KIVIBackbone=Llama-3.1-8B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.051
Block-GTQBackbone=Mistral-Nemo-12B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.0515
Block-GTQBackbone=Mean, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.054
Block-GTQBackbone=GLM-4-9B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0546
Block-GTQBackbone=DS-V2-Lite, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0548
KIVIBackbone=Qwen2.5-14B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0558
TQ-MSEBackbone=Qwen2.5-14B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0567
Block-GTQBackbone=Llama-3.1-8B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.0569
TQ-MSEBackbone=Qwen3-30B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0571
KIVIBackbone=DS-R1-Llama-8B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0584
TQ-MSEBackbone=Qwen3-8B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0594
KIVIBackbone=Llama-3.1-8B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0614
KIVIBackbone=DS-R1-Llama-8B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0663
KIVIBackbone=Qwen2.5-3B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.068
KIVIBackbone=DS-V2-Lite, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0684
KIVIBackbone=Qwen3-30B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0696
TQ-MSEBackbone=Mean, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0769
Block-GTQBackbone=Qwen2.5-14B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.0773
KIVIBackbone=GLM-4-9B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0829
TQ-MSEBackbone=Qwen2.5-3B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.0859
KIVIBackbone=Qwen2.5-3B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.0989
TQ-MSEBackbone=DS-R1-Llama-8B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.1024
TQ-MSEBackbone=GLM-4-9B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.1072
KIVIBackbone=GLM-4-9B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.1188
Block-GTQBackbone=Qwen3-8B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.121
Block-GTQBackbone=DS-R1-Qwen-7B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.1229
Block-GTQBackbone=Qwen3-30B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.1234
KIVIBackbone=DS-R1-Llama-8B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.1256
TQ-MSEBackbone=DS-V2-Lite, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.1274
TQ-MSEBackbone=Mistral-Nemo-12B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.1288
KIVIBackbone=Llama-3.1-8B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.1298
TQ-MSEBackbone=Llama-3.1-8B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.1444
KIVIBackbone=Qwen2.5-14B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.1596
Block-GTQBackbone=GLM-4-9B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.1617
Block-GTQBackbone=Qwen2.5-3B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.1636
Block-GTQBackbone=Mean, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.178
Block-GTQBackbone=DS-V2-Lite, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.1799
TQ-MSEBackbone=Qwen3-8B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.196
KIVIBackbone=Qwen3-30B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.202
KIVIBackbone=DS-V2-Lite, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.2023
TQ-MSEBackbone=Mean, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.2097
TQ-MSEBackbone=Qwen2.5-14B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.2118
TQ-MSEBackbone=Qwen3-30B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.2288
TQ-MSEBackbone=Qwen2.5-3B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.2544
TQ-MSEBackbone=DS-R1-Qwen-7B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.2915
KIVIBackbone=Mean, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.3053
KIVIBackbone=Qwen2.5-3B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.3114
KIVIBackbone=Mean, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.3275
KIVIBackbone=Qwen3-8B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.3338
TQ-MSEBackbone=GLM-4-9B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.345
KIVIBackbone=GLM-4-9B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.3492
KIVIBackbone=Qwen3-8B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.3594
TQ-MSEBackbone=DS-V2-Lite, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.4381
Block-GTQBackbone=GLM-4-9B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.4546
KIVIBackbone=Mean, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.4708
TQ-MSEBackbone=DS-R1-Qwen-7B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.484
KIVIBackbone=Qwen3-8B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.5126
KIVIBackbone=DS-R1-Qwen-7B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.060.5326
KIVIBackbone=DS-R1-Qwen-7B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.060.5569
Block-GTQBackbone=DS-V2-Lite, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.5709
TQ-MSEBackbone=Mean, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.5847
TQ-MSEBackbone=Qwen3-8B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.6349
TQ-MSEBackbone=Qwen2.5-3B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.6359
KIVIBackbone=DS-R1-Qwen-7B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.7481
TQ-MSEBackbone=Qwen3-30B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.8097
TQ-MSEBackbone=DS-R1-Qwen-7B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.8782
TQ-MSEBackbone=GLM-4-9B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.060.9397
TQ-MSEBackbone=DS-V2-Lite, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.061.3615
KIVIBackbone=Mistral-Nemo-12B, Bit Budget=4 b/dim, Buffer Setting=no-buffer2026.061.7932
KIVIBackbone=Mistral-Nemo-12B, Bit Budget=3 b/dim, Buffer Setting=no-buffer2026.061.8197
KIVIBackbone=Mistral-Nemo-12B, Bit Budget=2 b/dim, Buffer Setting=no-buffer2026.061.9675