train_config.json

{"comet": false, "comet_api_key": null, "comet_workspace": null, "comet_project": null, "comet_run_id": "fgp4puis", "wandb": false, "wandb_api_key": null, "wandb_entity": null, "wandb_project": null, "wandb_run_id": "fgp4puis", "base_model_dir": "/llm-downloader-destination/base/fireworks/deepseek-v3p1/hf", "output_model_dir": "gs://fireworks-artifacts-resendezcordell-oah2-bbb5dc/tuned-model-fgp4puis/ccce3b/ft-0121-dyck2-300/checkpoint", "checkpoint_dir": "/dev/shm/checkpoints", "gcs_checkpoint_dir": "gs://fireworks-artifacts-resendezcordell-oah2-bbb5dc/tuned-model-fgp4puis/ccce3b/ft-0121-dyck2-300/checkpoints/checkpoints", "max_checkpoints_to_keep": 1, "checkpoint_interval": 3600, "save_final_checkpoint": false, "train": true, "learning_rate": 0.0001, "learning_rate_warmup_steps": 0, "grad_accum_steps": 1, "epochs": 1, "early_stop": false, "seed": 42, "dataset_dir": "/mnt/staging/dataset", "eval_auto_carveout": true, "eval_dataset_dir": null, "train_limit": null, "max_context_len": 3200, "batch_size": 3200, "batch_size_samples": null, "max_data_workers": 0, "min_evals_per_epoch": 1, "max_evals_per_epoch": 5, "precision": null, "status_file": "gs://fireworks-fine-tuning-job-status/sftj-resendezcordell-oah2-fgp4puis-8a9346d0-bd7e-4e19-9f87-2d0eb438926f", "billing_file": "gs://fireworks-fine-tuning-metadata/sftj-resendezcordell-oah2-fgp4puis/billing-8a9346d0-bd7e-4e19-9f87-2d0eb438926f", "metrics_file": "gs://fireworks-fine-tuning-metadata/sftj-resendezcordell-oah2-fgp4puis/metrics.jsonl", "trainer_logs_file": null, "profile": null, "weight_sharding": null, "activation_sharding": null, "empty_weights": false, "nan_ratio_threshold": 0.05, "fast_api_port": 80, "optimizer": "adamw", "optimizer_weight_decay": 0.01, "target_shard_size_gb": null, "enable_fast_processor": false, "peft_addon_dir": null, "lora_rank": 8, "lora_dropout": 0.05, "template_kind": "conversation", "template": null, "mtp_config": {"enable_mtp": false, "freeze_base_model": false, "num_draft_tokens": 1}, "qat": true, "kld": false, "teft_tokens": [], "skip_dataset_filtering": false}