{
    "adapter_names": [
        "guardian-core",
        "decision"
    ],
    "adapter_ranks": [
        16,
        64
    ],
    "adapter_substitute_token_ids": [
        27,
        14711
    ],
    "adapter_token_ids": [
        100352,
        100353
    ],
    "architectures": [
        "GraniteSwitchForCausalLM"
    ],
    "asr_chunk_length_s": 120.0,
    "asr_chunk_overlap_s": 5.0,
    "asr_device": "cuda",
    "asr_dtype": null,
    "asr_enabled": true,
    "asr_generate_kwargs": null,
    "asr_max_audio_clips": 32,
    "asr_model_id": null,
    "asr_pipeline_kwargs": null,
    "asr_self_chunks": false,
    "attention_bias": false,
    "attention_dropout": 0.0,
    "attention_multiplier": 0.0078125,
    "bos_token_id": 100257,
    "control_token_gain": 15.0,
    "cross_stream_rank": null,
    "dtype": "bfloat16",
    "dual_stream": false,
    "embedding_multiplier": 12.0,
    "eos_token_id": 100257,
    "fused_add_norm": false,
    "hidden_act": "silu",
    "hidden_size": 4096,
    "initializer_range": 0.1,
    "intermediate_size": 12800,
    "layer_types": [
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention",
        "full_attention"
    ],
    "logits_scaling": 16.0,
    "lora_target_modules": [
        "o_proj",
        "qkv_proj",
        "shared_input_linear",
        "shared_output_linear"
    ],
    "mamba_chunk_size": 256,
    "mamba_conv_bias": true,
    "mamba_d_conv": 4,
    "mamba_d_head": 64,
    "mamba_d_state": 256,
    "mamba_expand": 2,
    "mamba_n_groups": 1,
    "mamba_n_heads": 128,
    "mamba_proj_bias": false,
    "max_lora_rank": 64,
    "max_position_embeddings": 131072,
    "model_type": "granite_switch",
    "ms_code_m": 6,
    "ms_code_type": "kerdock",
    "ms_counting_head_dim": 32,
    "ms_memory_gain": 28.0,
    "num_adapters": 2,
    "num_attention_heads": 32,
    "num_experts_per_tok": 2,
    "num_hidden_layers": 42,
    "num_key_value_heads": 8,
    "num_local_experts": 0,
    "output_router_logits": false,
    "pad_token_id": 100256,
    "position_embedding_type": "rope",
    "projection_head_dim": 128,
    "quantization": {
        "group_size": 64,
        "bits": 8,
        "mode": "affine"
    },
    "quantization_config": {
        "group_size": 64,
        "bits": 8,
        "mode": "affine"
    },
    "residual_multiplier": 0.22,
    "rms_norm_eps": 1e-05,
    "rope_parameters": {
        "rope_theta": 10000000,
        "rope_type": "default"
    },
    "router_aux_loss_coef": 0.001,
    "shared_intermediate_size": 12800,
    "switch_head_dim": 32,
    "tie_word_embeddings": true,
    "time_step_limit": [
        0.0,
        {
            "__float__": "Infinity"
        }
    ],
    "transformers_version": "5.17.0",
    "use_cache": true,
    "vocab_size": 100355
}