{ "_name_or_path": "fusion-8b-base", "architectures": ["FusionModel"], "model_type": "fusion", "vocab_size": 100000, "hidden_size": 4096, "num_hidden_layers": 32, "num_attention_heads": 32, "num_key_value_heads": 8, "intermediate_size": 11008, "hidden_act": "silu", "hidden_dropout_prob": 0.0, "attention_probs_dropout_prob": 0.0, "rms_norm_eps": 1e-6, "use_cache": true, "max_position_embeddings": 32768, "initializer_range": 0.02, "rope_theta": 10000.0, "rope_scaling": null, "attention_bias": false, "mlp_bias": false, "attention_dropout": 0.0, "block_size": 512, "latent_dim": 64, "window_size": 2048, "sbla_mode": "mixed", "enable_thinking_dial": true, "num_thinking_depths": 4, "torch_dtype": "bfloat16", "transformers_version": "4.36.0", "attn_implementation": "eager", "pad_token_id": 0, "bos_token_id": 1, "eos_token_id": 2, "tie_word_embeddings": false }