Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Arithmetic Reasoning on GSM8K (ACC, PPL)
Loading...
30.25
ACC
FP16 baseline
19.1324
22.0187
24.905
27.7913
Aug 13, 2023
ACC
PPL
Updated 4d ago
Evaluation Results
Method
Method
Links
ACC
PPL
FP16 baseline
Backbone Model=LLaMA-7B
2023.08
30.25
1.47
TSLD
Backbone Model=LLaMA-7B
2023.08
26.23
1.52
Logit KD
Backbone Model=LLaMA-7B
2023.08
25.47
1.52
TSLD
Backbone Model=OPT-6.7B
2023.08
24.49
2.14
FP16 baseline
Backbone Model=OPT-6.7B
2023.08
22.52
1.89
Logit KD
Backbone Model=OPT-6.7B
2023.08
21.08
1.93
OPT-2.7B FP16
Model=OPT-2.7B, Precis...
2023.08
20.39
2.07
TSLD
Model=OPT-2.7B, Quanti...
2023.08
20.24
2.03
Logit
Model=OPT-2.7B, Quanti...
2023.08
20.02
2.03
GT+Logit
Model=OPT-2.7B, Quanti...
2023.08
19.56
2.12
Feedback
Search any
task
Search any
task