See axolotl config

axolotl version: 0.4.1

yaml

1adapter: lora 2base_model: hf-tiny-model-private/tiny-random-OPTForCausalLM 3bf16: true 4chat_template: llama3 5dataloader_num_workers: 6 6dataset_prepared_path: null 7datasets: 8- data_files: 9 - 9742221b6e81649f_train_data.json 10 ds_type: json 11 format: custom 12 path: /workspace/input_data/9742221b6e81649f_train_data.json 13 type: 14 field_input: language 15 field_instruction: company 16 field_output: sentence 17 format: '{instruction} {input}' 18 no_input_format: '{instruction}' 19 system_format: '{system}' 20 system_prompt: '' 21debug: null 22deepspeed: null 23early_stopping: 24 metric: eval_loss 25 mode: min 26 patience: 3 27eval_max_new_tokens: 128 28eval_steps: 200 29eval_table_size: null 30evals_per_epoch: null 31flash_attention: true 32fp16: true 33fsdp: null 34fsdp_config: null 35gradient_accumulation_steps: 4 36gradient_checkpointing: false 37group_by_length: true 38hub_model_id: error577/942cfd35-534e-4ab7-9dd2-335ce292d7c2 39hub_repo: null 40hub_strategy: checkpoint 41hub_token: null 42learning_rate: 0.0002 43load_in_4bit: false 44load_in_8bit: false 45local_rank: null 46logging_steps: 1 47lora_alpha: 16 48lora_dropout: 0.3 49lora_fan_in_fan_out: null 50lora_model_dir: null 51lora_r: 8 52lora_target_linear: true 53lr_scheduler: cosine 54max_grad_norm: 1.0 55max_steps: 3500 56micro_batch_size: 16 57mlflow_experiment_name: /tmp/9742221b6e81649f_train_data.json 58model_type: AutoModelForCausalLM 59num_epochs: 50 60optimizer: adamw_bnb_8bit 61output_dir: miner_id_24 62pad_to_sequence_len: true 63resume_from_checkpoint: null 64s2_attention: null 65sample_packing: false 66save_steps: 200 67sequence_len: 64 68strict: false 69tf32: false 70tokenizer_type: AutoTokenizer 71train_on_inputs: false 72trust_remote_code: true 73val_set_size: 0.02 74wandb_entity: null 75wandb_mode: online 76wandb_name: 1be7f53b-7c51-4d01-9396-e086bf3f09dd 77wandb_project: Gradients-On-Demand 78wandb_run: your_name 79wandb_runid: 1be7f53b-7c51-4d01-9396-e086bf3f09dd 80warmup_steps: 10 81weight_decay: 0.01 82xformers_attention: null 83

942cfd35-534e-4ab7-9dd2-335ce292d7c2

• Loss: 6.9015

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

• learning_rate: 0.0002
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• training_steps: 3500

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