{ "architectures": [ "AsteriskForCausalLM" ], "aspp_dropout": 0.1, "aspp_hidden_dim": 256, "aspp_num_neighbors": 1, "aspp_num_steps": 4, "attention_bias": false, "attention_dropout": 0.0, "auto_map": { "AutoConfig": "AsteriskForCausalLM.AsteriskConfig", "AutoModelForCausalLM": "AsteriskForCausalLM.AsteriskForCausalLM" }, "bos_token_id": 1, "dtype": "bfloat16", "eos_token_id": 2, "head_dim": 64, "hidden_act": "silu", "hidden_size": 576, "hybrid_layer_indices": null, "initializer_range": 0.041666666666666664, "intermediate_size": 1536, "is_llama_config": true, "max_position_embeddings": 8192, "mlp_bias": false, "model_type": "asterisk", "num_attention_heads": 9, "num_hidden_layers": 30, "num_key_value_heads": 3, "pad_token_id": 2, "pi_flow": true, "pi_flow_scale": 1.0, "pi_flow_steps": 2, "pi_flow_use_gate": true, "pretraining_tp": 1, "rms_norm_eps": 1e-05, "rope_interleaved": false, "rope_scaling": null, "rope_theta": 100000, "tie_word_embeddings": true, "transformers.js_config": { "kv_cache_dtype": { "fp16": "float16", "q4f16": "float16" } }, "transformers_version": "4.57.6", "use_cache": true, "vocab_size": 49152 }