| { |
| "architectures": [ |
| "Lfm2MoeForCausalLM" |
| ], |
| "bos_token_id": null, |
| "conv_L_cache": 3, |
| "conv_bias": false, |
| "dtype": "bfloat16", |
| "eos_token_id": 6002, |
| "hidden_size": 1024, |
| "initializer_range": 0.02, |
| "intermediate_size": 256, |
| "layer_types": [ |
| "conv", |
| "conv", |
| "full_attention", |
| "conv", |
| "conv", |
| "conv", |
| "full_attention", |
| "conv", |
| "conv", |
| "conv", |
| "full_attention", |
| "conv" |
| ], |
| "max_position_embeddings": 128000, |
| "model_type": "lfm2_moe", |
| "moe_intermediate_size": 128, |
| "norm_eps": 1e-05, |
| "norm_topk_prob": true, |
| "num_attention_heads": 32, |
| "num_dense_layers": 2, |
| "num_experts": 8, |
| "num_experts_per_tok": 2, |
| "num_hidden_layers": 12, |
| "num_key_value_heads": 8, |
| "pad_token_id": 1, |
| "rope_parameters": { |
| "rope_theta": 1000000.0, |
| "rope_type": "default" |
| }, |
| "routed_scaling_factor": 1.0, |
| "tie_word_embeddings": true, |
| "transformers_version": "5.3.0", |
| "use_cache": false, |
| "use_expert_bias": true, |
| "vocab_size": 6026 |
| } |
|
|
