config.json

{
  "architectures": [
    "JetNemotronForCausalLM"
  ],
  "attention_dropout": 0.0,
  "auto_map": {
    "AutoConfig": "configuration_jet_nemotron.JetNemotronConfig",
    "AutoModelForCausalLM": "modeling_jet_nemotron.JetNemotronForCausalLM"
  },
  "bos_token_id": 151643,
  "efficient_attention_config": {
    "jet": {
      "conv_size": 4,
      "dconv_generator_reduction": 8,
      "dconv_implementation": "triton",
      "expand_v": 2,
      "head_dim": 128,
      "mode": "chunk",
      "norm_eps": "1e-5",
      "num_heads": 16
    },
    "swa": {
      "window_size": 2048
    }
  },
  "eos_token_id": 151643,
  "hidden_act": "silu",
  "hidden_size": 2048,
  "initializer_range": 0.02,
  "intermediate_size": 11008,
  "layer_types": [
    "jet",
    "jet",
    "jet",
    "jet",
    "jet",
    "swa",
    "jet",
    "jet",
    "jet",
    "jet",
    "jet",
    "jet",
    "jet",
    "jet",
    "jet",
    "jet",
    "swa",
    "attn",
    "jet",
    "swa",
    "attn",
    "swa",
    "swa",
    "jet",
    "jet",
    "swa",
    "jet",
    "swa",
    "jet",
    "jet",
    "jet",
    "jet",
    "attn",
    "jet",
    "jet",
    "jet"
  ],
  "max_position_embeddings": 32768,
  "max_window_layers": 36,
  "model_type": "jet_nemotron",
  "num_attention_heads": 16,
  "num_hidden_layers": 36,
  "num_key_value_heads": 2,
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 1000000.0,
  "tie_word_embeddings": true,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.51.3",
  "use_cache": true,
  "use_mrope": false,
  "vocab_size": 151936
}