curious-rwr
This model was trained from scratch on the kejian/codeparrot-train-more-filter-3.3b-cleaned dataset.
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0005
- train_batch_size: 64
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.01
- training_steps: 12588
- mixed_precision_training: Native AMP
Framework versions
- Transformers 4.23.0
- Pytorch 1.13.0+cu116
- Datasets 2.0.0
- Tokenizers 0.12.1
Full config
{'dataset': {'datasets': ['kejian/codeparrot-train-more-filter-3.3b-cleaned'],
'is_split_by_sentences': True,
'skip_tokens': 1649999872},
'generation': {'batch_size': 128,
'every_n_steps': 512,
'force_call_on': [12588],
'metrics_configs': [{}, {'n': 1}, {}],
'scenario_configs': [{'display_as_html': True,
'generate_kwargs': {'do_sample': True,
'eos_token_id': 0,
'max_length': 640,
'min_length': 10,
'temperature': 0.7,
'top_k': 0,
'top_p': 0.9},
'name': 'unconditional',
'num_hits_threshold': 0,
'num_samples': 2048},
{'display_as_html': True,
'generate_kwargs': {'do_sample': True,
'eos_token_id': 0,
'max_length': 272,
'min_length': 10,
'temperature': 0.7,
'top_k': 0,
'top_p': 0.9},
'name': 'functions',
'num_hits_threshold': 0,
'num_samples': 2048,
'prompts_path': 'resources/functions_csnet.jsonl',
'use_prompt_for_scoring': True}],
'scorer_config': {}},
'kl_gpt3_callback': {'every_n_steps': 512,
'force_call_on': [12588],
'gpt3_kwargs': {'model_name': 'code-cushman-001'},
'max_tokens': 64,
'num_samples': 4096},
'model': {'from_scratch': False,
'gpt2_config_kwargs': {'reorder_and_upcast_attn': True,
'scale_attn_by': True},
'model_kwargs': {'revision': 'c38e2b6acf17781918d39a310ee1adc4674a8225',
'value_head_config': {'is_detached': False}},
'path_or_name': 'kejian/mighty-rwr'},
'objective': {'alpha': 1, 'beta': 10, 'name': 'AWR'},
'tokenizer': {'path_or_name': 'codeparrot/codeparrot-small'},
'training': {'dataloader_num_workers': 0,
'effective_batch_size': 128,
'evaluation_strategy': 'no',
'fp16': True,
'hub_model_id': 'curious-rwr',
'hub_strategy': 'all_checkpoints',
'learning_rate': 0.0005,
'logging_first_step': True,
'logging_steps': 1,
'num_tokens': 3300000000.0,
'output_dir': 'training_output',
'per_device_train_batch_size': 16,
'push_to_hub': True,
'remove_unused_columns': False,
'save_steps': 12588,
'save_strategy': 'steps',
'seed': 42,
'tokens_already_seen': 1649999872,
'warmup_ratio': 0.01,
'weight_decay': 0.1}}
Wandb URL:
