| { | |
| "architectures": [ | |
| "Qwen3NextForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 0, | |
| "decoder_sparse_step": 1, | |
| "dtype": "float32", | |
| "eos_token_id": 0, | |
| "full_attention_interval": 3, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 128, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 256, | |
| "layer_types": [ | |
| "linear_attention", | |
| "linear_attention", | |
| "full_attention", | |
| "linear_attention", | |
| "linear_attention", | |
| "full_attention" | |
| ], | |
| "linear_conv_kernel_dim": 2, | |
| "linear_key_head_dim": 32, | |
| "linear_num_key_heads": 4, | |
| "linear_num_value_heads": 8, | |
| "linear_value_head_dim": 32, | |
| "max_position_embeddings": 2048, | |
| "mlp_only_layers": [], | |
| "model_type": "qwen3_next", | |
| "moe_intermediate_size": 64, | |
| "norm_topk_prob": true, | |
| "num_attention_heads": 2, | |
| "num_experts": 128, | |
| "num_experts_per_tok": 4, | |
| "num_hidden_layers": 6, | |
| "num_key_value_heads": 1, | |
| "output_router_logits": false, | |
| "partial_rotary_factor": 0.25, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "router_aux_loss_coef": 0.001, | |
| "shared_expert_intermediate_size": 256, | |
| "tie_word_embeddings": true, | |
| "transformers_version": "4.57.0", | |
| "use_cache": true, | |
| "vocab_size": 151936 | |
| } | |