jsingh's picture
Upload LoRA checkpoint: Qwen3-32B verifier adapter (rank=64, alpha=128)
02d89bf verified
{
"epoch": 2.0,
"total_flos": 1679413557067776.0,
"train_loss": 0.07952569609889067,
"train_runtime": 91366.2212,
"train_samples_per_second": 0.484,
"train_steps_per_second": 0.061
}