| { |
| "activation_function": "silu", |
| "architectures": [ |
| "GraniteMoeForCausalLM" |
| ], |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "attention_multiplier": 0.015625, |
| "bos_token_id": 100257, |
| "dtype": "float32", |
| "embedding_multiplier": 12, |
| "eos_token_id": 100257, |
| "hidden_act": "silu", |
| "hidden_size": 1024, |
| "initializer_range": 0.1, |
| "intermediate_size": 2048, |
| "logits_scaling": 4, |
| "max_position_embeddings": 32768, |
| "model_type": "granitemoe", |
| "num_attention_heads": 16, |
| "num_experts_per_tok": 4, |
| "num_hidden_layers": 28, |
| "num_key_value_heads": 4, |
| "num_local_experts": 8, |
| "output_router_logits": false, |
| "pad_token_id": 100256, |
| "position_embedding_type": "rope", |
| "residual_multiplier": 0.263, |
| "rms_norm_eps": 1e-05, |
| "rope_scaling": null, |
| "rope_theta": 10000000, |
| "router_aux_loss_coef": 0.01, |
| "router_jitter_noise": 0.0, |
| "tie_word_embeddings": true, |
| "transformers_version": "4.57.1", |
| "use_cache": true, |
| "vocab_size": 100352 |
| } |
|
|