CoolFace
Modelpublic

genies-models/openllama-3b-code_low_quality

sourceHugging Faceupdated 3y agoView on Hugging Face
0likes3downloads
train_args.json123 linesDownload Raw Back to root
1{2    "initial_model_dir": "models/openllama-3b",3    "distribution_id": "code_low_quality",4    "date_trained": "20/10/2023 00:16:48",5    "output_dir": "models/classify_lora/openllama-3b-code_low_quality",6    "overwrite_output_dir": false,7    "do_train": false,8    "do_eval": true,9    "do_predict": false,10    "evaluation_strategy": "steps",11    "prediction_loss_only": false,12    "per_device_train_batch_size": 16,13    "per_device_eval_batch_size": 32,14    "per_gpu_train_batch_size": null,15    "per_gpu_eval_batch_size": null,16    "gradient_accumulation_steps": 1,17    "eval_accumulation_steps": null,18    "eval_delay": 0,19    "learning_rate": 7.2e-05,20    "weight_decay": 0.0,21    "adam_beta1": 0.9,22    "adam_beta2": 0.999,23    "adam_epsilon": 1e-08,24    "max_grad_norm": 0.3,25    "num_train_epochs": 3,26    "max_steps": 100,27    "lr_scheduler_type": "constant",28    "warmup_ratio": 0.03,29    "warmup_steps": 0,30    "log_level": "passive",31    "log_level_replica": "warning",32    "log_on_each_node": true,33    "logging_dir": "models/classify_lora/openllama-3b-code_low_quality/runs/Oct19_23-15-39_compute-permanent-node-978",34    "logging_strategy": "steps",35    "logging_first_step": false,36    "logging_steps": 1,37    "logging_nan_inf_filter": true,38    "save_strategy": "steps",39    "save_steps": 25,40    "save_total_limit": 0,41    "save_safetensors": false,42    "save_on_each_node": false,43    "no_cuda": false,44    "use_cpu": false,45    "use_mps_device": false,46    "seed": 42,47    "data_seed": null,48    "jit_mode_eval": false,49    "use_ipex": false,50    "bf16": false,51    "fp16": false,52    "fp16_opt_level": "O1",53    "half_precision_backend": "auto",54    "bf16_full_eval": false,55    "fp16_full_eval": false,56    "tf32": null,57    "local_rank": 0,58    "ddp_backend": null,59    "tpu_num_cores": null,60    "tpu_metrics_debug": false,61    "debug": [],62    "dataloader_drop_last": false,63    "eval_steps": 25,64    "dataloader_num_workers": 0,65    "past_index": -1,66    "run_name": "train|models-classify_lora-openllama-3b-code_low_quality",67    "disable_tqdm": false,68    "remove_unused_columns": false,69    "label_names": null,70    "load_best_model_at_end": false,71    "metric_for_best_model": "eval_code_low_quality_score",72    "greater_is_better": true,73    "ignore_data_skip": false,74    "sharded_ddp": [],75    "fsdp": [],76    "fsdp_min_num_params": 0,77    "fsdp_config": {78        "min_num_params": 0,79        "xla": false,80        "xla_fsdp_grad_ckpt": false81    },82    "fsdp_transformer_layer_cls_to_wrap": null,83    "deepspeed": "configs/ds_zero_1.json",84    "label_smoothing_factor": 0.0,85    "optim": "paged_adamw_32bit",86    "optim_args": null,87    "adafactor": false,88    "group_by_length": false,89    "length_column_name": "length",90    "report_to": [91        "wandb"92    ],93    "ddp_find_unused_parameters": false,94    "ddp_bucket_cap_mb": null,95    "ddp_broadcast_buffers": null,96    "dataloader_pin_memory": true,97    "skip_memory_metrics": true,98    "use_legacy_prediction_loop": false,99    "push_to_hub": false,100    "resume_from_checkpoint": null,101    "hub_model_id": null,102    "hub_strategy": "every_save",103    "hub_token": null,104    "hub_private_repo": false,105    "hub_always_push": false,106    "gradient_checkpointing": false,107    "include_inputs_for_metrics": false,108    "fp16_backend": "auto",109    "push_to_hub_model_id": null,110    "push_to_hub_organization": null,111    "push_to_hub_token": null,112    "_n_gpu": 1,113    "mp_parameters": "",114    "auto_find_batch_size": false,115    "full_determinism": false,116    "torchdynamo": null,117    "ray_scope": "last",118    "ddp_timeout": 1800,119    "torch_compile": false,120    "torch_compile_backend": null,121    "torch_compile_mode": null,122    "dispatch_batches": null123}