tiny-random/lfm2-moe
0528
1{2 "architectures": [3 "Lfm2MoeForCausalLM"4 ],5 "bos_token_id": 1,6 "conv_L_cache": 3,7 "conv_bias": false,8 "dtype": "bfloat16",9 "eos_token_id": 7,10 "hidden_size": 64,11 "intermediate_size": 128,12 "layer_types": [13 "conv",14 "conv",15 "full_attention"16 ],17 "max_position_embeddings": 128000,18 "model_type": "lfm2_moe",19 "moe_intermediate_size": 128,20 "norm_eps": 1e-05,21 "norm_topk_prob": true,22 "num_attention_heads": 2,23 "num_dense_layers": 2,24 "num_experts": 32,25 "num_experts_per_tok": 4,26 "num_hidden_layers": 3,27 "num_key_value_heads": 1,28 "pad_token_id": 0,29 "rope_parameters": {30 "rope_theta": 1000000.0,31 "rope_type": "default"32 },33 "rope_theta": 1000000.0,34 "routed_scaling_factor": 1.0,35 "transformers_version": "5.0.0.dev0",36 "use_cache": true,37 "use_expert_bias": true,38 "vocab_size": 6553639}40 