| { | |
| "absolute_position_embedding": false, | |
| "architectures": [ | |
| "MotifForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "attn_rms_norm_eps": 1e-05, | |
| "auto_map": { | |
| "AutoConfig": "configuration_motif.MotifConfig", | |
| "AutoModelForCausalLM": "modeling_motif.MotifForCausalLM" | |
| }, | |
| "bfloat16": true, | |
| "bos_token_id": 219396, | |
| "dtype": "float32", | |
| "eos_token_id": 219395, | |
| "expanded": true, | |
| "fused_rope": false, | |
| "head_dim": 128, | |
| "hidden_act": "poly_norm", | |
| "hidden_size": 4096, | |
| "initializer_range": 2e-05, | |
| "intermediate_size": 16384, | |
| "k_ratio": 1, | |
| "load_pretrained": null, | |
| "loss_reduction": "mean", | |
| "max_position_embeddings": 131072, | |
| "max_window_layers": 28, | |
| "model_type": "Motif", | |
| "num_attention_heads": 40, | |
| "num_hidden_layers": 40, | |
| "num_key_value_heads": 16, | |
| "num_noise_heads": 8, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": { | |
| "factor": 2.0, | |
| "original_max_position_embeddings": 65536, | |
| "rope_type": "yarn" | |
| }, | |
| "rope_theta": 2000000.0, | |
| "sliding_window": null, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "4.57.1", | |
| "use_bias": false, | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "vocab_size": 219520 | |
| } | |