| !!python/tuple | |
| - !!python/object:__main__.ModelArguments | |
| bnb_4bit_quant_type: nf4 | |
| cache_dir: ./cache | |
| device_map: auto | |
| load_in_4bit: true | |
| load_in_8bit: false | |
| model_name_or_path: HuggingFaceH4/zephyr-7b-beta | |
| model_revision: main | |
| model_type: auto | |
| neft_alpha: 0 | |
| rope_scaling: null | |
| shift_attn: false | |
| tokenizer_name_or_path: null | |
| torch_dtype: float16 | |
| trust_remote_code: true | |
| use_bnb_nested_quant: false | |
| use_fast_tokenizer: false | |
| use_flash_attention_2: false | |
| - !!python/object:__main__.DataArguments | |
| dataset_config_name: null | |
| dataset_name: null | |
| ignore_pad_token_for_loss: true | |
| max_eval_samples: null | |
| max_train_samples: null | |
| overwrite_cache: false | |
| preprocessing_num_workers: 4 | |
| template_name: vicuna | |
| train_file_dir: datasets/finetune | |
| validation_file_dir: null | |
| validation_split_percentage: 10 | |
| - !!python/object:__main__.SFTConfig | |
| __cached__setup_devices: !!python/object/apply:torch.device | |
| - cuda | |
| - 0 | |
| _n_gpu: 1 | |
| adafactor: false | |
| adam_beta1: 0.9 | |
| adam_beta2: 0.999 | |
| adam_epsilon: 1.0e-08 | |
| auto_find_batch_size: false | |
| bf16: false | |
| bf16_full_eval: false | |
| data_seed: null | |
| dataloader_drop_last: false | |
| dataloader_num_workers: 0 | |
| dataloader_pin_memory: true | |
| ddp_backend: null | |
| ddp_broadcast_buffers: null | |
| ddp_bucket_cap_mb: null | |
| ddp_find_unused_parameters: false | |
| ddp_timeout: 30000 | |
| debug: [] | |
| deepspeed: null | |
| deepspeed_plugin: null | |
| disable_tqdm: false | |
| dispatch_batches: null | |
| distributed_state: !!python/object:accelerate.state.PartialState | |
| _cpu: false | |
| backend: null | |
| debug: false | |
| device: !!python/object/apply:torch.device | |
| - cuda | |
| - 0 | |
| distributed_type: !!python/object/apply:accelerate.utils.dataclasses.DistributedType | |
| - MULTI_GPU | |
| fork_launched: false | |
| local_process_index: 0 | |
| num_processes: 1 | |
| process_index: 0 | |
| do_eval: true | |
| do_predict: false | |
| do_train: true | |
| eval_accumulation_steps: null | |
| eval_delay: 0 | |
| eval_steps: 25 | |
| evaluation_strategy: !!python/object/apply:transformers.trainer_utils.IntervalStrategy | |
| - steps | |
| fp16: false | |
| fp16_backend: auto | |
| fp16_full_eval: false | |
| fp16_opt_level: O1 | |
| fsdp: [] | |
| fsdp_config: | |
| min_num_params: 0 | |
| xla: false | |
| xla_fsdp_grad_ckpt: false | |
| fsdp_min_num_params: 0 | |
| fsdp_transformer_layer_cls_to_wrap: null | |
| full_determinism: false | |
| gradient_accumulation_steps: 1 | |
| gradient_checkpointing: true | |
| gradient_checkpointing_kwargs: | |
| use_reentrant: false | |
| greater_is_better: null | |
| group_by_length: false | |
| half_precision_backend: auto | |
| hub_always_push: false | |
| hub_model_id: hllj/non-qa-sft-zephyr-7b-beta-v1 | |
| hub_private_repo: false | |
| hub_strategy: !!python/object/apply:transformers.trainer_utils.HubStrategy | |
| - every_save | |
| hub_token: null | |
| ignore_data_skip: false | |
| include_inputs_for_metrics: false | |
| include_tokens_per_second: false | |
| jit_mode_eval: false | |
| label_names: null | |
| label_smoothing_factor: 0.0 | |
| learning_rate: 3.0e-05 | |
| length_column_name: length | |
| load_best_model_at_end: false | |
| local_rank: 0 | |
| log_level: info | |
| log_level_replica: warning | |
| log_on_each_node: true | |
| logging_dir: outputs-sft-zephyr-beta-v1/runs/Nov22_05-52-29_a72e59c0abac | |
| logging_first_step: true | |
| logging_nan_inf_filter: true | |
| logging_steps: 10 | |
| logging_strategy: !!python/object/apply:transformers.trainer_utils.IntervalStrategy | |
| - steps | |
| lr_scheduler_type: !!python/object/apply:transformers.trainer_utils.SchedulerType | |
| - cosine | |
| max_grad_norm: 1.0 | |
| max_seq_length: 512 | |
| max_steps: 50 | |
| metric_for_best_model: null | |
| mp_parameters: '' | |
| neftune_noise_alpha: null | |
| no_cuda: false | |
| num_train_epochs: 3.0 | |
| optim: !!python/object/apply:transformers.training_args.OptimizerNames | |
| - adamw_torch | |
| optim_args: null | |
| output_dir: outputs-sft-zephyr-beta-v1 | |
| overwrite_output_dir: true | |
| past_index: -1 | |
| per_device_eval_batch_size: 4 | |
| per_device_train_batch_size: 4 | |
| per_gpu_eval_batch_size: null | |
| per_gpu_train_batch_size: null | |
| prediction_loss_only: false | |
| push_to_hub: true | |
| push_to_hub_model_id: null | |
| push_to_hub_organization: null | |
| push_to_hub_token: null | |
| ray_scope: last | |
| remove_unused_columns: true | |
| report_to: | |
| - wandb | |
| resume_from_checkpoint: null | |
| run_name: sft-zephyr-7b-beta-v1 | |
| save_on_each_node: false | |
| save_safetensors: true | |
| save_steps: 25 | |
| save_strategy: !!python/object/apply:transformers.trainer_utils.IntervalStrategy | |
| - steps | |
| save_total_limit: 13 | |
| seed: 42 | |
| skip_memory_metrics: true | |
| split_batches: false | |
| tf32: null | |
| torch_compile: false | |
| torch_compile_backend: null | |
| torch_compile_mode: null | |
| torchdynamo: null | |
| tpu_metrics_debug: false | |
| tpu_num_cores: null | |
| use_cpu: false | |
| use_ipex: false | |
| use_legacy_prediction_loop: false | |
| use_mps_device: false | |
| warmup_ratio: 0.05 | |
| warmup_steps: 0 | |
| weight_decay: 0.05 | |
| - !!python/object:__main__.ScriptArguments | |
| lora_alpha: 16 | |
| lora_dropout: 0.1 | |
| lora_modules_to_save: null | |
| lora_r: 64 | |
| lora_target_modules: | |
| - q_proj | |
| - k_proj | |
| - v_proj | |
| - o_proj | |
| peft_path: null | |
| use_peft: true | |