Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Language Modeling on Open Australian Legal Corpus
Loading...
1.32
Loss
Original
1.2892
1.4971
1.705
1.9129
Feb 11, 2026
Loss
PPL
Updated 1mo ago
Evaluation Results
Method
Method
Links
Loss
PPL
Original
Backbone=Llama-2-7b
2026.02
1.32
3.76
Pruning
Backbone=Llama-2-7b
2026.02
1.32
3.75
Ghost
Backbone=Llama-2-7b
2026.02
1.7
5.45
Pruning
Backbone=Llama-3-8b
2026.02
1.73
5.65
Pruning
Backbone=GPT-2
2026.02
1.74
5.71
Original
Backbone=Llama-3-8b
2026.02
1.74
5.72
Pruning
Backbone=Llama-3.1-8b
2026.02
1.74
5.71
Original
Backbone=Llama-3.1-8b
2026.02
1.75
5.76
Original
Backbone=GPT-2
2026.02
1.78
5.91
Ghost
Backbone=GPT-2
2026.02
2.01
7.46
Ghost
Backbone=Llama-3-8b
2026.02
2.09
8.12
Ghost
Backbone=Llama-3.1-8b
2026.02
2.09
8.08
Feedback
Search any
task
Search any
task
