config.json

{
  "_name_or_path": "ibm-granite/granite-3.1-3b-a800m-instruct",
  "architectures": [
    "GraniteMoeForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "attention_multiplier": 0.015625,
  "bos_token_id": 0,
  "embedding_multiplier": 12.0,
  "eos_token_id": 0,
  "hidden_act": "silu",
  "hidden_size": 1536,
  "initializer_range": 0.02,
  "intermediate_size": 512,
  "logits_scaling": 6.0,
  "max_position_embeddings": 131072,
  "model_type": "granitemoe",
  "num_attention_heads": 24,
  "num_experts_per_tok": 8,
  "num_hidden_layers": 32,
  "num_key_value_heads": 8,
  "num_local_experts": 40,
  "output_router_logits": false,
  "pad_token_id": 0,
  "quantization_config": {
    "config_groups": {
      "group_0": {
        "input_activations": null,
        "output_activations": null,
        "targets": [
          "Linear"
        ],
        "weights": {
          "actorder": "weight",
          "block_structure": null,
          "dynamic": false,
          "group_size": 128,
          "num_bits": 4,
          "observer": "mse",
          "observer_kwargs": {},
          "strategy": "group",
          "symmetric": true,
          "type": "int"
        }
      }
    },
    "format": "pack-quantized",
    "global_compression_ratio": 1.4210061815911108,
    "ignore": [
      "model.layers.0.block_sparse_moe.router.layer",
      "model.layers.1.block_sparse_moe.router.layer",
      "model.layers.2.block_sparse_moe.router.layer",
      "model.layers.3.block_sparse_moe.router.layer",
      "model.layers.4.block_sparse_moe.router.layer",
      "model.layers.5.block_sparse_moe.router.layer",
      "model.layers.6.block_sparse_moe.router.layer",
      "model.layers.7.block_sparse_moe.router.layer",
      "model.layers.8.block_sparse_moe.router.layer",
      "model.layers.9.block_sparse_moe.router.layer",
      "model.layers.10.block_sparse_moe.router.layer",
      "model.layers.11.block_sparse_moe.router.layer",
      "model.layers.12.block_sparse_moe.router.layer",
      "model.layers.13.block_sparse_moe.router.layer",
      "model.layers.14.block_sparse_moe.router.layer",
      "model.layers.15.block_sparse_moe.router.layer",
      "model.layers.16.block_sparse_moe.router.layer",
      "model.layers.17.block_sparse_moe.router.layer",
      "model.layers.18.block_sparse_moe.router.layer",
      "model.layers.19.block_sparse_moe.router.layer",
      "model.layers.20.block_sparse_moe.router.layer",
      "model.layers.21.block_sparse_moe.router.layer",
      "model.layers.22.block_sparse_moe.router.layer",
      "model.layers.23.block_sparse_moe.router.layer",
      "model.layers.24.block_sparse_moe.router.layer",
      "model.layers.25.block_sparse_moe.router.layer",
      "model.layers.26.block_sparse_moe.router.layer",
      "model.layers.27.block_sparse_moe.router.layer",
      "model.layers.28.block_sparse_moe.router.layer",
      "model.layers.29.block_sparse_moe.router.layer",
      "model.layers.30.block_sparse_moe.router.layer",
      "model.layers.31.block_sparse_moe.router.layer",
      "lm_head"
    ],
    "kv_cache_scheme": null,
    "quant_method": "compressed-tensors",
    "quantization_status": "compressed"
  },
  "residual_multiplier": 0.22,
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 10000000.0,
  "router_aux_loss_coef": 0.0,
  "tie_word_embeddings": true,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.47.1",
  "use_cache": true,
  "vocab_size": 49155
}