ODE Discovery on PKPD
0.015RMSELLM-SR†
Evaluation results
| Method | Links | |
|---|---|---|
| LLM-SR†Access Level=Full trajectory access, LLM Backbone=GPT-5.22026.07 | 0.015 | |
| AgentODEAccess Level=Summary-level only, LLM Backbone=GPT-5.22026.07 | 0.064 | |
| AgentODE w/o iterative refinementAccess Level=Summary-level only, LLM Backbone=GPT-5.2, Refinement=None2026.07 | 0.259 | |
| LLM-based parameter initialization (k = 5)Access Level=Summary-level only, k=5, LLM Backbone=GPT-5.22026.07 | 0.53 | |
| Neural ODEAccess Level=Full trajectory access2026.07 | 0.685 | |
| PySRAccess Level=Full trajectory access2026.07 | 0.981 | |
| SINDyAccess Level=Full trajectory access2026.07 | 1.251 |