Model (HuggingFace ID or local path)
Dataset (HuggingFace ID)
Eval Dataset (optional HF ID)
Eval Split
Split
Max Samples
Precision
Max Seq Length
Epochs
Learning Rate
Batch Size
Grad Accum Steps
⚡ Efficiency
▼
🔗 LoRA / PEFT
▼
GEKO Config
▼
Advanced
▼
Output Directory
Resume from Checkpoint (optional — leave blank to start fresh)
▶ Start Training
■ Stop
✕ Clear
Status
Epoch
Step
Compute Saved
Latest Loss
Steps/sec
ETA
Eval Loss
Loss Curve
Bucket Distribution
Output Directory