| _config_type: | |
| _is_type: true | |
| _module: capsules.train | |
| _qualname: TrainConfig | |
| cache_ema_alpha: 0.9 | |
| context: | |
| _config_type: | |
| _is_type: true | |
| _module: capsules.tasks.longhealth.context | |
| _qualname: LongHealthStructuredContextConfig | |
| patient_ids: | |
| - patient_01 | |
| - patient_02 | |
| - patient_03 | |
| - patient_04 | |
| - patient_05 | |
| - patient_06 | |
| - patient_07 | |
| - patient_08 | |
| - patient_09 | |
| - patient_10 | |
| dataset: | |
| _config_type: | |
| _is_type: true | |
| _module: capsules.datasets | |
| _qualname: CapsuleDatasetLatest.Config | |
| convo_transforms: null | |
| data_sources: | |
| - !!python/tuple | |
| - hazy-research/capsules/generate_longhealth_simple_p10_s5_n65536:v0 | |
| - null | |
| - !!python/tuple | |
| - hazy-research/capsules/generate_longhealth_simple_p10_s5_n65536:v1 | |
| - null | |
| dataset_weights: null | |
| is_wandb: true | |
| kwargs: {} | |
| label_type: logits | |
| max_sequence_length: 1024 | |
| target: | |
| _is_type: true | |
| _module: capsules.datasets | |
| _qualname: CapsuleDatasetLatest | |
| top_k_logits: 20 | |
| user_prompt_prefix: null | |
| device: cuda | |
| distributed_backend: gloo | |
| ema_cache: false | |
| epochs: 2 | |
| eval_datasets: | |
| - _config_type: | |
| _is_type: true | |
| _module: capsules.train | |
| _qualname: EvalDatasetConfig | |
| dataloader_num_workers: 0 | |
| dataset: | |
| _config_type: | |
| _is_type: true | |
| _module: capsules.tasks.longhealth | |
| _qualname: LongHealthEvalDataset.Config | |
| convo_transforms: null | |
| data_sources: [] | |
| dataset_weights: null | |
| is_wandb: false | |
| kwargs: {} | |
| label_type: tokens | |
| max_questions: 256 | |
| patient_ids: | |
| - patient_01 | |
| - patient_02 | |
| - patient_03 | |
| - patient_04 | |
| - patient_05 | |
| - patient_06 | |
| - patient_07 | |
| - patient_08 | |
| - patient_09 | |
| - patient_10 | |
| target: | |
| _is_type: true | |
| _module: capsules.tasks.longhealth | |
| _qualname: LongHealthEvalDataset | |
| top_k_logits: 20 | |
| user_prompt_prefix: null | |
| local_batch_size: 16 | |
| name_for_wandb: longhealth_mc | |
| only_eval_rank_0: false | |
| eval_every_n_steps: 256 | |
| eval_log_table: true | |
| eval_max_samples: null | |
| generate_datasets: | |
| - _config_type: | |
| _is_type: true | |
| _module: capsules.train | |
| _qualname: GenerateDatasetConfig | |
| batch_size: 16 | |
| dataloader_num_workers: 0 | |
| dataset: | |
| _config_type: | |
| _is_type: true | |
| _module: capsules.tasks.longhealth | |
| _qualname: LongHealthMultipleChoiceGenerateDataset.Config | |
| cot: true | |
| include_diagnosis: true | |
| kwargs: {} | |
| max_questions: null | |
| patient_ids: | |
| - patient_01 | |
| - patient_02 | |
| - patient_03 | |
| - patient_04 | |
| - patient_05 | |
| - patient_06 | |
| - patient_07 | |
| - patient_08 | |
| - patient_09 | |
| - patient_10 | |
| target: | |
| _is_type: true | |
| _module: capsules.tasks.longhealth | |
| _qualname: LongHealthMultipleChoiceGenerateDataset | |
| name_for_wandb: longhealth_mc | |
| num_samples: 4 | |
| num_samples_final: 8 | |
| override_max_tokens: null | |
| temperature: 0.3 | |
| generate_every_n_steps: 512 | |
| generate_max_new_tokens: 512 | |
| global_batch_size: 64 | |
| keep_last_n_saved: 1 | |
| kv_cache_initializer: | |
| _config_type: | |
| _is_type: true | |
| _module: capsules.kv_initialization.strategies.first_n_tokens | |
| _qualname: KVCacheInitFromFirstNTokensOfContext.Config | |
| context: null | |
| kwargs: {} | |
| max_tokens: 2048 | |
| num_frozen_tokens: 1 | |
| target: | |
| _is_type: true | |
| _module: capsules.kv_initialization.strategies.first_n_tokens | |
| _qualname: KVCacheInitFromFirstNTokensOfContext | |
| launch_id: 2025-05-10-14-56-42-train_longhealth_simple | |
| local_batch_size: 4 | |
| log_logprob_viz: false | |
| loss_type: logits | |
| lr: 0.02 | |
| lr_scheduler: null | |
| max_optimizer_steps: -1 | |
| model: | |
| _config_type: | |
| _is_type: true | |
| _module: capsules.config | |
| _qualname: HFModelConfig | |
| attn_implementation: einsum | |
| checkpoint_path: null | |
| load_kwargs: {} | |
| model_cls: | |
| _is_type: true | |
| _module: capsules.models.llama | |
| _qualname: LlamaForCausalLM | |
| peft: | |
| _config_type: | |
| _is_type: true | |
| _module: capsules.config | |
| _qualname: PeftConfig | |
| adapter_non_linearity: relu | |
| adapter_reduction_factor: 16 | |
| alpha: 16 | |
| bias: none | |
| dropout: 0.0 | |
| enabled: false | |
| encoder_dropout: 0.0 | |
| encoder_hidden_size: null | |
| encoder_reparameterization_type: MLP | |
| extra_params: {} | |
| method: lora | |
| num_virtual_tokens: 20 | |
| prefix_projection: false | |
| prompt_tuning_init: null | |
| prompt_tuning_init_text: null | |
| r: 8 | |
| target_modules: null | |
| task_type: CAUSAL_LM | |
| pretrained_model_name_or_path: meta-llama/Llama-3.2-3B-Instruct | |
| tuning_method: custom_prefix | |
| name: train_longhealth_simple_p10_lr0.02_toks2048 | |
| online_model: true | |
| optimizer: adam | |
| output_dir: /data/sabri/capsules | |
| pretrained_cache_path: null | |
| run_dir: /data/sabri/capsules/2025-05-10-14-56-42-train_longhealth_simple/68e4c064-dc5a-46c8-a726-b3c7977e9e1a | |
| run_id: 68e4c064-dc5a-46c8-a726-b3c7977e9e1a | |
| save_after_training: true | |
| save_every_n_steps: 512 | |
| save_to_wandb: true | |
| script_id: train_longhealth_simple | |
| seed: 42 | |
| tokenizer: meta-llama/Llama-3.2-1B-Instruct | |
| use_batch_sampler: false | |
| wandb: | |
| _config_type: | |
| _is_type: true | |
| _module: capsules.utils.wandb | |
| _qualname: WandBConfig | |
| entity: hazy-research | |
| group: null | |
| name: train_longhealth_simple_p10_lr0.02_toks2048 | |
| notes: null | |
| project: capsules | |
| tags: | |
| - train | |
| - longhealth | |
| - patientsp10 | |