{ "vocab_size": 32000, "d_model": 512, "n_layer": 12, "n_head": 8, "n_kv_head": 2, "d_ff": 1408, "max_seq_len": 1024, "rope_base": 10000.0, "rms_eps": 1e-05, "dropout": 0.0, "decision_head_classes": 5, "decision_head_dropout": 0.0, "mtp_k": 4, "mtp_head_kind": "mlp", "init_std": 0.02, "tie_word_embeddings": true, "ffn_kind": "swiglu", "ffn_rank": 128, "ffn_experts": 4, "ffn_top_k": 1, "chart_patch_encoder_enabled": false, "chart_image_size": 224, "chart_patch_size": 32, "chart_channels": 3, "chart_embed_dropout": 0.0 }