| { | |
| "architectures": [ | |
| "GraniteForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "attention_multiplier": 0.088388347648318, | |
| "bos_token_id": 128000, | |
| "dtype": "float32", | |
| "embedding_multiplier": 1.0, | |
| "eos_token_id": 128001, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 3072, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 8192, | |
| "logits_scaling": 55.425625842204, | |
| "max_position_embeddings": 4096, | |
| "mlp_bias": false, | |
| "model_type": "granite", | |
| "num_attention_heads": 24, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 6, | |
| "pad_token_id": 128001, | |
| "pretraining_tp": 1, | |
| "residual_multiplier": 1.0, | |
| "rms_norm_eps": 1e-05, | |
| "rope_parameters": { | |
| "rope_theta": 500000.0, | |
| "rope_type": "default" | |
| }, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "5.0.0", | |
| "use_cache": true, | |
| "vocab_size": 128256 | |
| } |