πŸŽ“

Training Data

Convert your high-quality predictions into SynLlama JSONL fine-tuning datasets β€” ready to ship to a cloud GPU
Loading stats…
🎯 Build Dataset
0.70
Higher = fewer but better examples
β€” examples will be exported
Refining…
πŸ“‹ Format Preview
SynLlama JSONL training format (Alpaca-style):
{
  "instruction": "You are an expert synthetic
   organic chemist...",
  "input": "Provide a synthetic pathway for this
   SMILES string: CC(=O)Oc1ccccc1C(=O)O",
  "output": "{\"reactions\":[...],
   \"building_blocks\":[...]}"
}
πŸ—ƒοΈ Past Exports
πŸ”¬ Matching Candidates β€” preview of what will be exported