| name: gpt2_text_generation | |
| config_type: model | |
| add_cross_attention: false | |
| vocab_size: 42001 | |
| attn_pdrop: 0.1 | |
| bos_token_id: 5 | |
| embd_pdrop: 0.1 | |
| eos_token_id: 5 | |
| gradient_checkpointing: false | |
| initializer_range: 0.02 | |
| layer_norm_epsilon: 1.0e-05 | |
| model_type: gpt2 | |
| n_ctx: 1024 | |
| n_embd: 768 | |
| n_head: 12 | |
| n_layer: 12 | |
| n_positions: 1024 | |
| resid_pdrop: 0.1 | |
| summary_activation: false | |
| summary_first_dropout: 0.1 | |
| use_cache: true | |
| generation: | |
| name: null | |
| config_type: model | |
| bos_token_id: 0 | |
| decoder_start_token_id: 0 | |
| early_stopping: true | |
| eos_token_id: 2 | |
| length_penalty: 2.0 | |
| max_new_tokens: 50 | |
| no_repeat_ngram_size: 3 | |
| num_beams: 4 | |
| pad_token_id: 1 | |