Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Language Modeling on LLaMA-2-7B
Loading...
5.47
Perplexity
BF16
-23.4372
171.6864
366.81
561.9336
Feb 16, 2024
Jun 21, 2024
Oct 26, 2024
Mar 2, 2025
Jul 7, 2025
Nov 11, 2025
Mar 18, 2026
Perplexity
Updated 1mo ago
Evaluation Results
Method
Method
Links
Perplexity
BF16
Quantization Bits=None...
2024.02
5.47
FP16
Type=Upper bound, Size...
2026.03
5.51
RAMP
Type=Mixed-precision,...
2026.03
5.54
AWQ-4
Type=SOTA, Size (GB)=3.90
2026.03
5.6
Q4_K_M (GGUF)
Type=Deployment, Size...
2026.03
5.61
QUIP#-4
Type=SOTA, Size (GB)=3.90
2026.03
5.62
GPTQ-4
Type=SOTA, Size (GB)=3.90
2026.03
5.69
RTN-4
Type=Naive, Size (GB)=...
2026.03
5.94
AQLM + PV
Variant=2-bit-1x16, Mo...
2026.02
6.08
AQLM + PV
Variant=2-bit-2x8, Mod...
2026.02
6.27
QTIP
Variant=2-bit, Model S...
2026.02
6.29
AQLM
Variant=2-bit-1x16, Mo...
2026.02
6.34
AQLM
Variant=2-bit-2x8, Mod...
2026.02
7.24
NANOQUANT
Variant=2-bit, Model S...
2026.02
7.35
BitDistiller
Quantization Bits=2, G...
2024.02
8.08
QuIP#
Quantization Bits=2, G...
2024.02
8.97
NANOQUANT
Variant=1-bit, Model S...
2026.02
9.01
QuIP
Quantization Bits=2, G...
2024.02
728.15
Feedback
Search any
task
Search any
task