最优模型训练参数

[jiechuangu]

我用如下参数进行训练，lora相关参数与3.5、模型权重中huggingface提供的相同，在Spider official size 95M database上evaluate，execution accuracy为0.766，使用huggingface提供的权重execution accuracy为0.787，请问可以提供huggingface上公开模型的训练参数吗？ model_args*************************** ModelArguments(model_name_or_path='codellama/CodeLlama-13b-Instruct-hf', cache_dir=None, use_fast_tokenizer=False, use_auth_token=False, model_revision='main', padding_side='left', quantization_bit=None, quantization_type='nf4', double_quantization=True, rope_scaling=None, checkpoint_dir=None, plot_loss=True, hf_auth_token=None, compute_dtype=torch.bfloat16, model_max_length=2560, hf_hub_token=None, split_special_tokens=False)

data_args*************************** DataArguments(template='llama2', dataset='example_text2sql_train', dataset_dir='dbgpt_hub/data/', cutoff_len=1024, reserved_label_len=1, split='train', streaming=False, buffer_size=16384, mix_strategy='concat', interleave_probs=None, overwrite_cache=True, preprocessing_num_workers=None, max_source_length=2048, max_target_length=512, max_samples=None, eval_num_beams=None, ignore_pad_token_for_loss=True, system_prompt=None, val_size=0, predicted_input_filename='dbgpt_hub/data/example_text2sql_dev.json', predicted_out_filename='pred_sql.sql')

training_args*************************** Seq2SeqTrainingArguments( _n_gpu=1, adafactor=False, adam_beta1=0.9, adam_beta2=0.999, adam_epsilon=1e-08, auto_find_batch_size=False, bf16=True, bf16_full_eval=False, data_seed=None, dataloader_drop_last=False, dataloader_num_workers=0, dataloader_pin_memory=True, ddp_backend=None, ddp_broadcast_buffers=None, ddp_bucket_cap_mb=None, ddp_find_unused_parameters=False, ddp_timeout=1800, debug=[], deepspeed=None, disable_tqdm=False, dispatch_batches=None, do_eval=False, do_predict=False, do_train=True, eval_accumulation_steps=None, eval_delay=0, eval_steps=None, evaluation_strategy=no, 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=None, full_determinism=False, generation_config=None, generation_max_length=None, generation_num_beams=None, gradient_accumulation_steps=16, gradient_checkpointing=False, greater_is_better=None, group_by_length=False, half_precision_backend=auto, hub_always_push=False, hub_model_id=None, hub_private_repo=False, hub_strategy=every_save, hub_token=<HUB_TOKEN>, ignore_data_skip=False, include_inputs_for_metrics=False, include_tokens_per_second=False, jit_mode_eval=False, label_names=None, label_smoothing_factor=0.0, learning_rate=0.0002, length_column_name=length, load_best_model_at_end=False, local_rank=0, log_level=passive, log_level_replica=warning, log_on_each_node=True, logging_dir=dbgpt_hub/output/adapter/CodeLlama-13b-sql-lora-test/runs/Apr15_08-36-28_cni-2288H-V5, logging_first_step=False, logging_nan_inf_filter=True, logging_steps=50, logging_strategy=steps, lr_scheduler_type=cosine_with_restarts, max_grad_norm=1.0, max_steps=-1, metric_for_best_model=None, mp_parameters=, no_cuda=False, num_train_epochs=8.0, optim=adamw_torch, optim_args=None, output_dir=dbgpt_hub/output/adapter/CodeLlama-13b-sql-lora-test, overwrite_output_dir=True, past_index=-1, per_device_eval_batch_size=8, per_device_train_batch_size=1, predict_with_generate=False, prediction_loss_only=False, push_to_hub=False, push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=<PUSH_TO_HUB_TOKEN>, ray_scope=last, remove_unused_columns=True, report_to=['wandb'], resume_from_checkpoint=None, run_name=dbgpt_hub/output/adapter/CodeLlama-13b-sql-lora-test, save_on_each_node=False, save_safetensors=False, save_steps=2000, save_strategy=steps, save_total_limit=None, seed=42, sharded_ddp=[], skip_memory_metrics=True, sortish_sampler=False, tf32=None, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, torchdynamo=None, tpu_metrics_debug=False, tpu_num_cores=None, use_cpu=False, use_ipex=False, use_legacy_prediction_loop=False, use_mps_device=False, warmup_ratio=0.0, warmup_steps=0, weight_decay=0.0, )

finetuning_args*************************** FinetuningArguments(stage='sft', finetuning_type='lora', num_hidden_layers=32, num_layer_trainable=3, name_module_trainable='mlp', lora_rank=64, lora_alpha=32.0, lora_dropout=0.1, lora_target=['q_proj', 'v_proj'], resume_lora_training=True, ppo_score_norm=False, dpo_beta=0.1)

generating_args*************************** GeneratingArguments(do_sample=True, temperature=0.95, top_p=0.7, top_k=50, num_beams=1, max_length=None, max_new_tokens=512, repetition_penalty=1.0, length_penalty=1.0)