Text Generation on P2 Prose prompt after Python training phase v1 (Evaluation)
5Coherent Emission RateTFGN GPT-2S
Evaluation results
| Method | Links | |
|---|---|---|
| TFGN GPT-2SModel Scale=398 M, TFGN Overlay=true, Tokenizer=GPT-2 BPE2026.05 | 5 | |
| TFGN GPT-2M FS+RFModel Scale=739 M × 2, Training Regime=From-Scratch + Retrofit, TFGN Overlay=true, Tokenizer=GPT-2 BPE2026.05 | 5 | |
| TFGN LLaMA 8B FS+RFModel Scale=9 B × 2, Training Regime=From-Scratch + Retrofit, TFGN Overlay=true, Tokenizer=LLaMA 3.1 8B2026.05 | 5 | |
| Std-FT GPT-2MModel Scale=GPT-2M, Fine-tuning Method=Std-FT, TFGN Overlay=false, Tokenizer=GPT-2 BPE2026.05 | 3 | |
| LoRA-r256 GPT-2MModel Scale=GPT-2M, Fine-tuning Method=LoRA-r256, TFGN Overlay=false, Tokenizer=GPT-2 BPE2026.05 | 3 | |
| Std-FT LLaMA 8BModel Scale=LLaMA 8B, Fine-tuning Method=Std-FT, TFGN Overlay=false, Tokenizer=LLaMA 3.1 8B2026.05 | 3 |