| { | |
| "_rope_scaling": null, | |
| "architectures": [ | |
| "Imu1ForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_imu_1.Imu1Config", | |
| "AutoModelForCausalLM": "modeling_imu_1.Imu1ForCausalLM" | |
| }, | |
| "attn_gating": "per-head", | |
| "attn_qknorm": true, | |
| "attn_val_residual": true, | |
| "bos_token_id": 0, | |
| "dtype": "float32", | |
| "eos_token_id": 0, | |
| "hidden_act": "silu", | |
| "hidden_size": 1152, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 3072, | |
| "layernorm_scaling": true, | |
| "max_position_embeddings": 8192, | |
| "model_type": "imu_1", | |
| "num_attention_heads": 18, | |
| "num_hidden_layers": 30, | |
| "num_key_value_heads": 6, | |
| "pad_token_id": 0, | |
| "rms_norm_eps": 1e-05, | |
| "rope_parameters": { | |
| "rope_theta": 10000, | |
| "rope_type": "default" | |
| }, | |
| "rope_theta": 10000, | |
| "tie_word_embeddings": false, | |
| "tokenizer_class": "PreTrainedTokenizerFast", | |
| "transformers_version": "5.0.0.dev0", | |
| "use_cache": true, | |
| "vocab_size": 49152 | |
| } | |