| architecture: |
| backbone_dtype: bfloat16 |
| gradient_checkpointing: true |
| intermediate_dropout: 0.0 |
| pretrained: true |
| pretrained_weights: /home/mtsmash/Schreibtisch/h2o-llmstudio-eva-gpt/output/user/EvaGPT-German-2B-Q10/checkpoint.pth |
| 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: Assistentin |
| chatbot_author: MaTeLiX AI |
| chatbot_name: h2oGPT |
| data_sample: 1.0 |
| data_sample_choice: |
| - Train |
| - Validation |
| id_column: id |
| limit_chained_samples: true |
| mask_prompt_labels: true |
| mask_prompt_user_text_only: false |
| only_last_answer: false |
| parent_id_column: parent_id |
| personalize: false |
| prompt_column: |
| - Kontext |
| - Benutzer |
| prompt_column_separator: \n\n |
| system_column: system |
| text_answer_separator: <|Assistentin|> |
| text_prompt_start: <|Benutzer|> |
| text_system_start: <|System|> |
| train_dataframe: /home/mtsmash/Schreibtisch/h2o-llmstudio-eva-gpt/data/user/04-07-2026-11/04-07-2026-11.csv |
| train_text_column: false |
| validation_dataframe: /home/mtsmash/Schreibtisch/h2o-llmstudio-eva-gpt/data/user/04-07-2026-11/04-07-2026-11.csv |
| validation_size: 0.1 |
| validation_strategy: automatic |
| environment: |
| compile_model: false |
| deepspeed_allgather_bucket_size: 1000000 |
| deepspeed_method: ZeRO2 |
| 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' |
| huggingface_branch: main |
| mixed_precision: false |
| mixed_precision_dtype: bfloat16 |
| number_of_workers: 8 |
| seed: -1 |
| trust_remote_code: false |
| use_deepspeed: true |
| experiment_name: EvaGPT-German-2B-Q11.1 |
| llm_backbone: /home/mtsmash/Schreibtisch/h2o-llmstudio-eva-gpt/output/user/EvaGPT-German-2B-Q10/ |
| logging: |
| log_all_ranks: false |
| log_step_size: absolute |
| logger: None |
| neptune_project: '' |
| wandb_entity: '' |
| wandb_project: '' |
| output_directory: /home/mtsmash/Schreibtisch/h2o-llmstudio-eva-gpt/output/user/EvaGPT-German-2B-Q11.1/ |
| prediction: |
| batch_size_inference: 0 |
| do_sample: false |
| max_length_inference: 2336 |
| max_time: 0.0 |
| metric: Perplexity |
| metric_gpt_model: gpt-3.5-turbo-0301 |
| metric_gpt_template: general |
| min_length_inference: 2 |
| num_beams: 1 |
| num_history: 4 |
| repetition_penalty: 1.1 |
| stop_tokens: '' |
| temperature: 0.0 |
| top_k: 40 |
| top_p: 0.9 |
| problem_type: text_causal_language_modeling |
| tokenizer: |
| add_prompt_answer_tokens: true |
| long_sample_strategy: Sliding Window |
| max_length: 2528 |
| padding_quantile: 1.0 |
| sliding_window_overlap: 256 |
| tokenizer_kwargs: '{"use_fast": true, "add_prefix_space": false}' |
| training: |
| attention_implementation: sdpa |
| batch_size: 4 |
| differential_learning_rate: 1.0e-05 |
| differential_learning_rate_layers: [] |
| drop_last_batch: true |
| epochs: 3 |
| evaluate_before_training: false |
| evaluation_epochs: 1.0 |
| freeze_layers: [] |
| grad_accumulation: 8 |
| gradient_clip: 1.0 |
| learning_rate: 0.0002 |
| lora: false |
| lora_alpha: 24 |
| lora_dropout: 0.05 |
| lora_r: 16 |
| lora_target_modules: q_proj,k_proj,v_proj,o_proj,gate_proj,up_proj,down_proj |
| lora_unfreeze_layers: [] |
| loss_function: StableTokenCrossEntropy |
| min_learning_rate_ratio: 0.0 |
| optimizer: AdamW |
| save_checkpoint: last |
| schedule: LossAwareCosine |
| train_validation_data: true |
| use_dora: false |
| use_rslora: false |
| warmup_epochs: 0.0 |
| weight_decay: 0.01 |
|
|
