config.json

{
  "adapt_base_lr": 1.0,
  "architectures": [
    "TTTMLPForCausalLM"
  ],
  "bos_token_id": 151645,
  "chunk_size": 16,
  "conv_kernel": 4,
  "eos_token_id": 151645,
  "hidden_act": "silu",
  "hidden_size": 2048,
  "initializer_range": 0.02,
  "intermediate_size": 5504,
  "max_position_embeddings": 4096,
  "memory_depth": 2,
  "mini_batch_eps": 1e-06,
  "mlp_bias": false,
  "model_type": "ttt_mlp",
  "num_attention_heads": 32,
  "num_hidden_layers": 24,
  "pad_token_id": 151643,
  "pretraining_tp": 1,
  "qkv_conv": false,
  "rms_norm_eps": 1e-06,
  "rope_scaling": {
    "attention_factor": 0.5,
    "beta_fast": 32.0,
    "beta_slow": 1.0,
    "factor": 32.0,
    "original_max_position_embeddings": 4096,
    "rope_type": "yarn"
  },
  "rope_theta": 100000.0,
  "scan_checkpoint_group_size": 0,
  "torch_dtype": "float32",
  "transformers_version": "4.55.0.dev0",
  "use_cache": true,
  "vocab_size": 151936
}