Language Modeling on GPT-2 124M (test)
17.33PerplexityDonor
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DonorModel=GPT-2 (124M), Trainable %=17.1%2026.01 | 17.33 | 12.8 | — | — | |
| DonorTrainable %=17.1%2026.01 | 17.33 | 12.8 | — | — | |
| LoRAModel=GPT-2 (124M), Trainable %=1.29%2026.01 | 668.4 | 29.8 | 2.3 | — | |
| LoRATrainable %=1.29%2026.01 | 668.4 | 29.8 | — | — | |
| Zero-shotModel=GPT-2 (124M), Trainable %=0%2026.01 | 697.76 | — | — | — | |
| Zero-shotTrainable %=0%2026.01 | 697.76 | — | — | — | |
| Full fine-tuneModel=GPT-2 (124M), Trainable %=100%2026.01 | 1,352.05 | 24 | 1.9 | — | |
| Full fine-tuneTrainable %=100%2026.01 | 1,352.05 | 24 | — | — | |
| Top-KModel=GPT-2 (124M), Trainable %=48.1%2026.01 | 2,021.55 | 20.9 | 1.6 | — | |
| Top-KTrainable %=48.1%2026.01 | 2,021.55 | 20.9 | — | — |