config.json

{
    "architectures": [
        "PlamoForCausalLM"
    ],
    "attention_window_size": 2048,
    "auto_map": {
        "AutoConfig": "modeling_plamo.PlamoConfig",
        "AutoModelForCausalLM": "modeling_plamo.PlamoForCausalLM"
    },
    "bos_token_id": 1,
    "capacity_factor": 1.0,
    "eos_token_id": 2,
    "eval_attention_n_bit": null,
    "eval_mlp_n_bit": null,
    "expert_dropout": 0.0,
    "fp8_accum_dtype": "bfloat16",
    "group_size": 1024,
    "hidden_size": 2048,
    "hidden_size_per_head": 128,
    "image_feature_size": null,
    "image_proj_type": "linear",
    "image_token_id": null,
    "intermediate_size": 8192,
    "k_expert": null,
    "linear_type": "fp8",
    "mamba_chunk_size": 256,
    "mamba_d_conv": 4,
    "mamba_d_state": 64,
    "mamba_enabled": true,
    "mamba_num_heads": 32,
    "mamba_step": 2,
    "max_position_embeddings": 10485760,
    "model_type": "plamo",
    "n_expert": null,
    "num_attention_heads": 16,
    "num_hidden_layers": 16,
    "num_key_value_heads": 1,
    "rms_norm_eps": 1e-06,
    "shared_intermediate_size": null,
    "sliding_window": 2048,
    "sparse_intermediate_size": null,
    "sparse_step": null,
    "tokenizer_class": "PlamoTokenizer",
    "torch_dtype": "float32",
    "transformers_version": "4.44.2",
    "use_cache": true,
    "use_predefined_initial_state": false,
    "vocab_size": 100000
}