PrimeIntellect/INTELLECT-3-Base
0216
1{2 "architectures": [3 "Glm4MoeForCausalLM"4 ],5 "auto_map": {6 "AutoConfig": "configuration_glm4_moe.Glm4MoeConfig",7 "AutoModelForCausalLM": "modeling_glm4_moe.Glm4MoeForCausalLM",8 "AutoModel": "modeling_glm4_moe.Glm4MoeModel"9 },10 "attention_bias": true,11 "attention_dropout": 0.0,12 "pad_token_id": 151329,13 "eos_token_id": [14 151334,15 15132916 ],17 "head_dim": 128,18 "hidden_act": "silu",19 "hidden_size": 4096,20 "partial_rotary_factor": 0.5,21 "initializer_range": 0.02,22 "intermediate_size": 10944,23 "max_position_embeddings": 131072,24 "model_type": "glm4_moe",25 "moe_intermediate_size": 1408,26 "norm_topk_prob": true,27 "num_attention_heads": 96,28 "n_group": 1,29 "topk_group": 1,30 "n_routed_experts": 128,31 "n_shared_experts": 1,32 "routed_scaling_factor": 1.0,33 "num_experts_per_tok": 8,34 "first_k_dense_replace": 1,35 "num_hidden_layers": 46,36 "num_key_value_heads": 8,37 "rms_norm_eps": 1e-05,38 "rope_scaling": null,39 "rope_theta": 1000000,40 "num_nextn_predict_layers": 1,41 "tie_word_embeddings": false,42 "torch_dtype": "bfloat16",43 "transformers_version": "4.54.0",44 "use_cache": true,45 "use_qk_norm": false,46 "vocab_size": 15155247}