|
config: |
|
(): colpali_engine.trainer.colmodel_training.ColModelTrainingConfig |
|
output_dir: !path ../../../models/biqwen2-1e-256-newpad-nobuffer |
|
processor: |
|
(): colpali_engine.utils.transformers_wrappers.AllPurposeWrapper |
|
class_to_instanciate: !ext colpali_engine.models.BiQwen2Processor |
|
pretrained_model_name_or_path: "./models/colqwen2_base" |
|
|
|
|
|
model: |
|
(): colpali_engine.utils.transformers_wrappers.AllPurposeWrapper |
|
class_to_instanciate: !ext colpali_engine.models.BiQwen2 |
|
pretrained_model_name_or_path: "./models/colqwen2_base" |
|
torch_dtype: !ext torch.bfloat16 |
|
use_cache: false |
|
attn_implementation: "flash_attention_2" |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
dataset_loading_func: !ext colpali_engine.utils.dataset_transformation.load_train_set |
|
eval_dataset_loader: !import ../data/test_data.yaml |
|
|
|
|
|
run_eval: true |
|
add_suffix: true |
|
loss_func: |
|
(): colpali_engine.loss.bi_encoder_losses.BiPairwiseCELoss |
|
tr_args: |
|
(): transformers.training_args.TrainingArguments |
|
output_dir: null |
|
overwrite_output_dir: true |
|
num_train_epochs: 1 |
|
per_device_train_batch_size: 64 |
|
gradient_checkpointing: true |
|
gradient_checkpointing_kwargs: { "use_reentrant": false } |
|
|
|
|
|
per_device_eval_batch_size: 64 |
|
eval_strategy: "steps" |
|
dataloader_num_workers: 8 |
|
|
|
save_steps: 500 |
|
logging_steps: 10 |
|
eval_steps: 100 |
|
warmup_steps: 100 |
|
learning_rate: 5e-4 |
|
save_total_limit: 1 |
|
resume_from_checkpoint: false |
|
report_to: "wandb" |
|
|
|
|
|
peft_config: |
|
(): peft.LoraConfig |
|
r: 32 |
|
lora_alpha: 32 |
|
lora_dropout: 0.1 |
|
init_lora_weights: "gaussian" |
|
bias: "none" |
|
task_type: "FEATURE_EXTRACTION" |
|
target_modules: '(.*(model).*(down_proj|gate_proj|up_proj|k_proj|q_proj|v_proj|o_proj).*$)' |
|
|
|
|
|
|