Time-series forecasting on LLM traffic (test)
112,000,000MAEPhi + STM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Phi + STMModel=Phi, prompt_reprogramming=true, fine-tuning=false, num_return_sequences=1, temperature=0.12, top_p=0.9, do_sample=false2026.01 | 112,000,000 | 20,720,000,000,000,000 | 54.64 | 80.58 | |
| LLaMA + STMModel=LLaMA, prompt_reprogramming=true, fine-tuning=false, num_return_sequences=1, temperature=0.12, top_p=0.9, do_sample=false2026.01 | 112,000,000 | 20,720,000,000,000,000 | 69.73 | 88.47 | |
| Phi (Base)Model=Phi, prompt_reprogramming=false, fine-tuning=false, num_return_sequences=1, temperature=0.12, top_p=0.9, do_sample=false2026.01 | 248,000,000 | 106,700,000,000,000,000 | — | — | |
| Gemma + STMModel=Gemma, prompt_reprogramming=true, fine-tuning=false, num_return_sequences=1, temperature=0.12, top_p=0.9, do_sample=false2026.01 | 321,000,000 | 159,200,000,000,000,000 | 10.06 | 12.94 | |
| DeepSeek + STMModel=DeepSeek, prompt_reprogramming=true, fine-tuning=false, num_return_sequences=1, temperature=0.12, top_p=0.9, do_sample=false2026.01 | 322,000,000 | 133,500,000,000,000,000 | 10.35 | 19.96 | |
| Gemma (Base)Model=Gemma, prompt_reprogramming=false, fine-tuning=false, num_return_sequences=1, temperature=0.12, top_p=0.9, do_sample=false2026.01 | 357,000,000 | 182,900,000,000,000,000 | — | — | |
| DeepSeek (Base)Model=DeepSeek, prompt_reprogramming=false, fine-tuning=false, num_return_sequences=1, temperature=0.12, top_p=0.9, do_sample=false2026.01 | 359,000,000 | 166,800,000,000,000,000 | — | — | |
| LLaMA (Base)Model=LLaMA, prompt_reprogramming=false, fine-tuning=false, num_return_sequences=1, temperature=0.12, top_p=0.9, do_sample=false2026.01 | 370,000,000 | 179,700,000,000,000,000 | — | — |