| { | |
| "architectures": [ | |
| "GPTRefactForCausalLM" | |
| ], | |
| "attention_bias_in_fp32": true, | |
| "attention_softmax_in_fp32": true, | |
| "auto_map": { | |
| "AutoConfig": "configuration_gpt_refact.GPTRefactConfig", | |
| "AutoModelForCausalLM": "modeling_gpt_refact.GPTRefactForCausalLM" | |
| }, | |
| "do_sample": true, | |
| "eos_token_id": 0, | |
| "initializer_range": 0.02, | |
| "layer_norm_epsilon": 1e-05, | |
| "model_type": "gpt_refact", | |
| "multi_query": true, | |
| "n_embd": 2048, | |
| "n_head": 32, | |
| "n_inner": null, | |
| "n_layer": 32, | |
| "n_positions": 4096, | |
| "scale_attention_softmax_in_fp32": true, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.31.0", | |
| "use_cache": true, | |
| "vocab_size": 49216 | |
| } | |