{ "ablate": [], "architectures": [ "MiCRoOLMo" ], "attention_bias": false, "attention_dropout": 0.0, "backbone_num_layers": 16, "config_path": "configs/config_micro_olmo.yml", "eos_token_id": 100257, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "intermediate_size": 8192, "jitter_noise": 0, "loss_method": "all", "loss_type": "ForCausalLMLoss", "max_position_embeddings": 4096, "model_type": "olmo2", "num_attention_heads": 16, "num_experts": 4, "num_experts_per_tok": 1, "num_hidden_layers": 64, "num_key_value_heads": 16, "pad_token_id": 100277, "rms_norm_eps": 1e-06, "rope_scaling": null, "rope_theta": 500000, "tie_word_embeddings": false, "torch_dtype": "float32", "transformers_version": "4.51.3", "use_bfloat16": true, "use_cache": true, "use_router": true, "vocab_size": 100279 }