Cycle 0: New best model with eval_loss 0.666352 (Trained with Unsloth)

Best model so far - eval_loss: 0.9681 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 0.9615 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 0.9696 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 0.9375 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 0.9273 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 1.0512 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 0.9230 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 1.1201 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 1.1765 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 1.1666 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 0.9813 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 0.9535 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 0.9475 -------------------------------------------------------- (Trained with Unsloth)

Best model so far - eval_loss: 0.7707 -------------------------------------------------------- (Trained with Unsloth)