See axolotl config

axolotl version: 0.4.1

yaml

1adapter: lora 2base_model: lcw99/zephykor-ko-7b-chang 3bf16: true 4chat_template: llama3 5dataset_prepared_path: null 6datasets: 7- data_files: 8 - 42c742f85a540ce2_train_data.json 9 ds_type: json 10 format: custom 11 path: /workspace/input_data/42c742f85a540ce2_train_data.json 12 type: 13 field_instruction: prompt 14 field_output: solution 15 format: '{instruction}' 16 no_input_format: '{instruction}' 17 system_format: '{system}' 18 system_prompt: '' 19debug: null 20deepspeed: null 21device_map: auto 22do_eval: true 23early_stopping_patience: 2 24eval_batch_size: 4 25eval_max_new_tokens: 128 26eval_steps: 150 27eval_table_size: null 28evals_per_epoch: null 29flash_attention: true 30fp16: false 31fsdp: null 32fsdp_config: null 33gradient_accumulation_steps: 4 34gradient_checkpointing: true 35group_by_length: true 36hub_model_id: nttx/13a20cb6-0fc5-42da-9519-babbaa35b31f 37hub_repo: null 38hub_strategy: checkpoint 39hub_token: null 40learning_rate: 0.0002 41load_in_4bit: false 42load_in_8bit: false 43local_rank: null 44logging_steps: 50 45lora_alpha: 128 46lora_dropout: 0.3 47lora_fan_in_fan_out: null 48lora_model_dir: null 49lora_r: 64 50lora_target_linear: true 51lr_scheduler: cosine 52max_grad_norm: 1.0 53max_memory: 54 0: 75GB 55max_steps: 3000 56micro_batch_size: 4 57mlflow_experiment_name: /tmp/42c742f85a540ce2_train_data.json 58model_type: AutoModelForCausalLM 59num_epochs: 10 60optim_args: 61 adam_beta1: 0.9 62 adam_beta2: 0.95 63 adam_epsilon: 1e-5 64optimizer: adamw_bnb_8bit 65output_dir: miner_id_24 66pad_to_sequence_len: true 67resume_from_checkpoint: null 68s2_attention: null 69sample_packing: false 70save_steps: 150 71saves_per_epoch: null 72sequence_len: 512 73strict: false 74tf32: true 75tokenizer_type: AutoTokenizer 76train_on_inputs: false 77trust_remote_code: true 78val_set_size: 0.05 79wandb_entity: null 80wandb_mode: online 81wandb_name: e44975c3-72ab-4a77-9f65-64b36ebc90ca 82wandb_project: Gradients-On-Demand 83wandb_run: your_name 84wandb_runid: e44975c3-72ab-4a77-9f65-64b36ebc90ca 85warmup_steps: 50 86weight_decay: 0.0 87xformers_attention: null 88

13a20cb6-0fc5-42da-9519-babbaa35b31f

• Loss: 0.0717

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

• learning_rate: 0.0002
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=adam_beta1=0.9,adam_beta2=0.95,adam_epsilon=1e-5
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 50
• training_steps: 3000

Training results

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1