{ "architectures": [ "TinyMixtralForCausalLM" ], "attention_dropout": 0.0, "dtype": "float32", "expert_intermediate_size": 2389, "head_dim": 64, "hidden_size": 896, "initializer_range": 0.02, "max_position_embeddings": 2048, "model_type": "tinymixtral", "num_attention_heads": 14, "num_experts_per_tok": 2, "num_hidden_layers": 10, "num_key_value_heads": 2, "num_local_experts": 6, "rms_norm_eps": 1e-06, "rope_theta": 1000000.0, "router_aux_loss_coef": 0.01, "router_jitter_noise": 0.01, "transformers_version": "4.57.3", "vocab_size": 32000, "auto_map": { "AutoConfig": "configuration_tinymixtral.TinyMixtralConfig", "AutoModelForCausalLM": "modeling_tinymixtral.TinyMixtralForCausalLM" } }