| { | |
| "transformers_version": "5.5.3", | |
| "architectures": [ | |
| "Llama4ForCausalLM" | |
| ], | |
| "vocab_size": 128256, | |
| "hidden_size": 1152, | |
| "intermediate_size": 16384, | |
| "num_hidden_layers": 15, | |
| "num_attention_heads": 18, | |
| "num_key_value_heads": 6, | |
| "head_dim": 64, | |
| "hidden_act": "silu", | |
| "max_position_embeddings": 4096, | |
| "rms_norm_eps": 1e-05, | |
| "use_cache": true, | |
| "bos_token_id": 128000, | |
| "eos_token_id": [ | |
| 128001, | |
| 128008, | |
| 128009 | |
| ], | |
| "tie_word_embeddings": false, | |
| "attention_dropout": 0.0, | |
| "attention_bias": false, | |
| "model_type": "llama4_text" | |
| } |