qwen3-1.7.v0 / train_config.json
AltinAziziNovomind's picture
Upload 4 files
de81405 verified
{"comet": false, "comet_api_key": null, "comet_workspace": null, "comet_project": null, "comet_run_id": "qwen3-1p7b-only-placeholders", "wandb": false, "wandb_api_key": null, "wandb_entity": null, "wandb_project": null, "wandb_run_id": "qwen3-1p7b-only-placeholders", "base_model_dir": "/llm-downloader-destination/base/fireworks/qwen3-1p7b/hf", "output_model_dir": "gs://fireworks-artifacts-nm-bu-iagent-aa694e-577b15/sftj-qwen3-1p7b-only-placeholders/ed0b06/qwen3-1p7b-only-placeholders/checkpoint", "checkpoint_dir": "/dev/shm/checkpoints", "gcs_checkpoint_dir": "gs://fireworks-fine-tuning-checkpoints/sftj-nm-bu-iagent-aa694e-qwen3-1p7b-only-placeholders-d57f7241-225f-401e-8ff2-b22f6cfaa072/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": 3, "early_stop": false, "seed": 42, "dataset_dir": "/mnt/staging/dataset", "eval_auto_carveout": false, "eval_dataset_dir": "/mnt/staging/evaluation-dataset", "train_limit": null, "max_context_len": 8192, "batch_size": 32768, "max_data_workers": 0, "min_evals_per_epoch": 4, "precision": null, "status_file": "gs://fireworks-fine-tuning-job-status/sftj-nm-bu-iagent-aa694e-qwen3-1p7b-only-placeholders-d57f7241-225f-401e-8ff2-b22f6cfaa072", "billing_file": "gs://fireworks-fine-tuning-metadata/sftj-nm-bu-iagent-aa694e-qwen3-1p7b-only-placeholders/billing-d57f7241-225f-401e-8ff2-b22f6cfaa072", "metrics_file": "gs://fireworks-fine-tuning-metadata/sftj-nm-bu-iagent-aa694e-qwen3-1p7b-only-placeholders/metrics.jsonl", "profile": null, "weight_sharding": null, "activation_sharding": null, "empty_weights": false, "nan_ratio_threshold": 0.001, "fast_api_port": 80, "peft_addon_dir": null, "lora_rank": 16, "lora_dropout": 0.05, "template_kind": "conversation", "template": null, "eval_train_ratio": 0.02, "mtp_config": {"enable_mtp": false, "freeze_base_model": false, "num_draft_tokens": 1}, "qat": true, "kld": false, "teft_tokens": []}