architecture: backbone_dtype: bfloat16 force_embedding_gradients: false gradient_checkpointing: true intermediate_dropout: 0.0 pretrained: true pretrained_weights: '' augmentation: neftune_noise_alpha: 0.0 random_parent_probability: 0.0 skip_parent_probability: 0.0 token_mask_probability: 0.0 dataset: add_eos_token_to_answer: true add_eos_token_to_prompt: true add_eos_token_to_system: true answer_column: output chatbot_author: H2O.ai chatbot_name: h2oGPT data_sample: 1.0 data_sample_choice: - Train - Validation limit_chained_samples: true mask_prompt_labels: true parent_id_column: parent_id personalize: false prompt_column: - instruction system_column: None text_answer_separator: <|answer|> text_prompt_start: <|prompt|> text_system_start: <|system|> train_dataframe: sft.pq validation_dataframe: None validation_size: 0.01 validation_strategy: automatic environment: compile_model: false deepspeed_reduce_bucket_size: 1000000 deepspeed_stage3_param_persistence_threshold: 1000000 deepspeed_stage3_prefetch_bucket_size: 1000000 find_unused_parameters: false gpus: - '0' - '1' - '2' - '3' - '4' - '5' - '6' - '7' huggingface_branch: main mixed_precision: false number_of_workers: 8 seed: -1 trust_remote_code: true use_deepspeed: false experiment_name: h2o-danube-1.8b-sft llm_backbone: h2oai/h2o-danube-1.8b-base logging: logger: Neptune neptune_project: h2o/h2o output_directory: output/ prediction: batch_size_inference: 0 do_sample: false max_length_inference: 1023 metric: GPT metric_gpt_model: gpt-4-1106-preview metric_gpt_template: mt-bench min_length_inference: 2 num_beams: 1 num_history: 4 repetition_penalty: 1.1 stop_tokens: '' temperature: 0.0 top_k: 0 top_p: 1.0 problem_type: text_causal_language_modeling tokenizer: add_prefix_space: false add_prompt_answer_tokens: false max_length: 16384 max_length_answer: 8192 max_length_prompt: 8192 padding_quantile: 1.0 use_fast: true training: batch_size: 8 differential_learning_rate: 1.0e-05 differential_learning_rate_layers: [] drop_last_batch: true epochs: 1 evaluate_before_training: false evaluation_epochs: 0.25 grad_accumulation: 1 gradient_clip: 0.0 learning_rate: 5.0e-05 lora: false lora_alpha: 16 lora_dropout: 0.05 lora_r: 4 lora_target_modules: '' loss_function: TokenAveragedCrossEntropy optimizer: AdamW save_best_checkpoint: false schedule: Cosine train_validation_data: false use_flash_attention_2: true warmup_epochs: 0.05 weight_decay: 0.0