| { |
| "architectures": [ |
| "Qwen3_5MoeForCausalLM" |
| ], |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "bos_token_id": null, |
| "dtype": "bfloat16", |
| "eos_token_id": null, |
| "head_dim": 128, |
| "hidden_act": "silu", |
| "hidden_size": 256, |
| "initializer_range": 0.02, |
| "intermediate_size": 256, |
| "layer_types": [ |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "full_attention" |
| ], |
| "linear_conv_kernel_dim": 4, |
| "linear_key_head_dim": 128, |
| "linear_num_key_heads": 4, |
| "linear_num_value_heads": 8, |
| "linear_value_head_dim": 128, |
| "max_position_embeddings": 32768, |
| "model_type": "qwen3_5_moe_text", |
| "moe_intermediate_size": 256, |
| "num_attention_heads": 4, |
| "num_experts": 8, |
| "num_experts_per_tok": 2, |
| "num_hidden_layers": 4, |
| "num_key_value_heads": 2, |
| "output_router_logits": false, |
| "pad_token_id": null, |
| "partial_rotary_factor": 0.25, |
| "rms_norm_eps": 1e-06, |
| "rope_parameters": { |
| "mrope_interleaved": true, |
| "mrope_section": [ |
| 6, |
| 6, |
| 4 |
| ], |
| "partial_rotary_factor": 0.25, |
| "rope_theta": 10000000, |
| "rope_type": "default" |
| }, |
| "router_aux_loss_coef": 0.001, |
| "shared_expert_intermediate_size": 256, |
| "tie_word_embeddings": true, |
| "transformers_version": "5.16.0.dev0", |
| "use_cache": true, |
| "vocab_size": 248077 |
| } |
|
|
