config.json

{
  "adaptive_routing": true,
  "architectures": [
    "SlimMoEForCausalLM"
  ],
  "auto_map": {
    "AutoConfig": "configuration_slim_moe.SlimMoEConfig",
    "AutoModelForCausalLM": "modeling_slim_moe.SlimMoEForCausalLM"
  },
  "dim": 768,
  "dropout": 0.1,
  "hidden_dim": 1536,
  "max_seq_len": 2048,
  "model_type": "slim_moe",
  "num_experts": 4,
  "num_heads": 12,
  "num_hidden_layers": 16,
  "torch_dtype": "float32",
  "transformers_version": "4.53.2",
  "vocab_size": 50257
}