config.json

{
  "architectures": [
    "Lfm2ForCausalLM"
  ],
  "attention_bias": false,
  "block_auto_adjust_ff_dim": false,
  "block_ffn_dim_multiplier": 1.0,
  "block_multiple_of": 256,
  "bos_token_id": 0,
  "conv_L_cache": 3,
  "conv_bias": false,
  "conv_dim": 256,
  "conv_dim_out": 256,
  "conv_use_xavier_init": true,
  "dtype": "float32",
  "eos_token_id": 1,
  "head_dim": 64,
  "hidden_size": 256,
  "initializer_range": 0.02,
  "intermediate_size": 768,
  "layer_types": [
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv"
  ],
  "max_position_embeddings": 2048,
  "model_type": "lfm2",
  "norm_eps": 1e-05,
  "num_attention_heads": 4,
  "num_hidden_layers": 32,
  "num_key_value_heads": 4,
  "pad_token_id": 0,
  "rms_norm_eps": 1e-05,
  "rope_theta": 10000.0,
  "transformers_version": "4.57.1",
  "use_cache": true,
  "vocab_size": 256
}