Backup-bdg/Xoron-Dev-MultiMoe
843
1{2 "best_model_checkpoint": "/kaggle/working/xoron-final",3 "best_metric": 5.891898287038009,4 "epoch": 4,5 "epochs_completed": 4,6 "global_step": 72,7 "is_local_process_zero": true,8 "is_world_process_zero": true,9 "log_history": [],10 "logging_steps": 50,11 "max_steps": 72,12 "num_train_epochs": 4,13 "total_flos": 0,14 "train_batch_size": 1,15 "effective_batch_size": 16,16 "learning_rate": 0.0001,17 "max_grad_norm": 1.0,18 "trainable_components": [19 "llm",20 "cross_attention",21 "modality_markers"22 ],23 "frozen_components": [24 "vision",25 "video",26 "audio",27 "speech",28 "image_generation",29 "video_generation"30 ],31 "trial_name": null,32 "trial_params": null33}