| { | |
| "_remove_final_layer_norm": false, | |
| "activation_function": "relu", | |
| "alpha_pattern": {}, | |
| "architectures": [ | |
| "OPTForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "auto_mapping": { | |
| "base_model_class": "OPTForCausalLM", | |
| "parent_library": "transformers.models.opt.modeling_opt" | |
| }, | |
| "base_model_name_or_path": "facebook/galactica-125m", | |
| "bias": "none", | |
| "bos_token_id": 2, | |
| "do_layer_norm_before": true, | |
| "dropout": 0.1, | |
| "enable_bias": true, | |
| "eos_token_id": 2, | |
| "fan_in_fan_out": false, | |
| "ffn_dim": 3072, | |
| "hidden_size": 768, | |
| "inference_mode": true, | |
| "init_lora_weights": true, | |
| "init_std": 0.02, | |
| "layer_norm_elementwise_affine": true, | |
| "layer_replication": null, | |
| "layerdrop": 0.0, | |
| "layers_pattern": null, | |
| "layers_to_transform": null, | |
| "loftq_config": {}, | |
| "lora_alpha": 16, | |
| "lora_dropout": 0, | |
| "max_position_embeddings": 2048, | |
| "megatron_config": null, | |
| "megatron_core": "megatron.core", | |
| "model_type": "opt", | |
| "modules_to_save": null, | |
| "num_attention_heads": 12, | |
| "num_hidden_layers": 12, | |
| "pad_token_id": 1, | |
| "peft_type": "LORA", | |
| "r": 16, | |
| "rank_pattern": {}, | |
| "revision": null, | |
| "target_modules": [ | |
| "k_proj", | |
| "v_proj", | |
| "down_proj", | |
| "q_proj", | |
| "o_proj", | |
| "gate_proj", | |
| "up_proj" | |
| ], | |
| "task_type": null, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.41.2", | |
| "use_cache": true, | |
| "use_dora": false, | |
| "use_rslora": false, | |
| "vocab_size": 50272, | |
| "word_embed_proj_dim": 768 | |
| } | |