parameters.json.pub

{
    "force": true,
    "dump_path": "serialization_dir/codegpt_intrain_20-06-2023_13-58-49",
    "data_file": "data/binarized_text.gpt2.pickle",
    "student_config": "training_configs/codegpt.json",
    "student_pretrained_weights": "microsoft/CodeGPT-small-py-adaptedGPT2",
    "teacher_name": "microsoft/CodeGPT-small-py-adaptedGPT2",
    "temperature": 2.0,
    "alpha_ce": 0.33,
    "alpha_clm": 0.33,
    "alpha_mse": 0.0,
    "alpha_cos": 0.33,
    "freeze_pos_embs": true,
    "n_epoch": 1,
    "batch_size": 6,
    "group_by_size": true,
    "gradient_accumulation_steps": 50,
    "warmup_prop": 0.05,
    "weight_decay": 0.0,
    "learning_rate": 0.0005,
    "adam_epsilon": 1e-06,
    "max_grad_norm": 5.0,
    "initializer_range": 0.02,
    "fp16": false,
    "fp16_opt_level": "O1",
    "n_gpu": 1,
    "local_rank": 0,
    "seed": 56,
    "log_interval": 500,
    "checkpoint_interval": 3900,
    "teacher_not_pretrained": false,
    "n_nodes": 1,
    "node_id": 0,
    "global_rank": 0,
    "world_size": 1,
    "n_gpu_per_node": 1,
    "multi_gpu": false,
    "is_master": true,
    "multi_node": false
}