config.json

{
  "_name_or_path": "yaya-sy/minixtral",
  "architectures": [
          "MixtralForCausalLM"
  ],
  "auto_map": {
          "AutoModelForCausalLM": "modeling_mixtral.MixtralForCausalLM",
          "AutoConfig": "configuration_mixtral.MixtralConfig"
  },
  "attention_dropout": 0.0,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "hidden_act": "silu",
  "hidden_size": 4096,
  "initializer_range": 0.02,
  "intermediate_size": 14336,
  "max_position_embeddings": 32768,
  "model_type": "mixtral",
  "num_attention_heads": 32,
  "num_experts_per_tok": 2,
  "num_hidden_layers": 32,
  "num_key_value_heads": 8,
  "num_local_experts": 8,
  "output_router_logits": false,
  "ranks": {
    "model.layers.0.block_sparse_moe.experts.0.w1": 1024,
    "model.layers.0.block_sparse_moe.experts.0.w2": 1024,
    "model.layers.0.block_sparse_moe.experts.0.w3": 1024,
    "model.layers.0.block_sparse_moe.experts.1.w1": 1024,
    "model.layers.0.block_sparse_moe.experts.1.w2": 1024,
    "model.layers.0.block_sparse_moe.experts.1.w3": 1024,
    "model.layers.0.block_sparse_moe.experts.2.w1": 1024,
    "model.layers.0.block_sparse_moe.experts.2.w2": 1024,
    "model.layers.0.block_sparse_moe.experts.2.w3": 1024,
    "model.layers.0.block_sparse_moe.experts.3.w1": 1024,
    "model.layers.0.block_sparse_moe.experts.3.w2": 1024,
    "model.layers.0.block_sparse_moe.experts.3.w3": 1024,
    "model.layers.0.block_sparse_moe.experts.4.w1": 1024,
    "model.layers.0.block_sparse_moe.experts.4.w2": 1024,
    "model.layers.0.block_sparse_moe.experts.4.w3": 1024,
    "model.layers.0.block_sparse_moe.experts.5.w1": 1024,
    "model.layers.0.block_sparse_moe.experts.5.w2": 1024,
    "model.layers.0.block_sparse_moe.experts.5.w3": 1024,
    "model.layers.0.block_sparse_moe.experts.6.w1": 1024,
    "model.layers.0.block_sparse_moe.experts.6.w2": 1024,
    "model.layers.0.block_sparse_moe.experts.6.w3": 1024,
    "model.layers.0.block_sparse_moe.experts.7.w1": 1024,
    "model.layers.0.block_sparse_moe.experts.7.w2": 1024,
    "model.layers.0.block_sparse_moe.experts.7.w3": 1024,
    "model.layers.1.block_sparse_moe.experts.0.w1": 1024,
    "model.layers.1.block_sparse_moe.experts.0.w2": 1024,
    "model.layers.1.block_sparse_moe.experts.0.w3": 1024,
    "model.layers.1.block_sparse_moe.experts.1.w1": 1024,
    "model.layers.1.block_sparse_moe.experts.1.w2": 1024,
    "model.layers.1.block_sparse_moe.experts.1.w3": 1024,
    "model.layers.1.block_sparse_moe.experts.2.w1": 1024,
    "model.layers.1.block_sparse_moe.experts.2.w2": 1024,
    "model.layers.1.block_sparse_moe.experts.2.w3": 1024,
    "model.layers.1.block_sparse_moe.experts.3.w1": 1024,
    "model.layers.1.block_sparse_moe.experts.3.w2": 1024,
    "model.layers.1.block_sparse_moe.experts.3.w3": 1024,
    "model.layers.1.block_sparse_moe.experts.4.w1": 1024,
    "model.layers.1.block_sparse_moe.experts.4.w2": 1024,
    "model.layers.1.block_sparse_moe.experts.4.w3": 1024,
    "model.layers.1.block_sparse_moe.experts.5.w1": 1024,
    "model.layers.1.block_sparse_moe.experts.5.w2": 1024,
    "model.layers.1.block_sparse_moe.experts.5.w3": 1024,
    "model.layers.1.block_sparse_moe.experts.6.w1": 1024,
    "model.layers.1.block_sparse_moe.experts.6.w2": 1024,
    "model.layers.1.block_sparse_moe.experts.6.w3": 1024,
    "model.layers.1.block_sparse_moe.experts.7.w1": 1024,
    "model.layers.1.block_sparse_moe.experts.7.w2": 1024,
    "model.layers.1.block_sparse_moe.experts.7.w3": 1024,
    "model.layers.2.block_sparse_moe.experts.0.w1": 1024,
    "model.layers.2.block_sparse_moe.experts.0.w2": 1024,
    "model.layers.2.block_sparse_moe.experts.0.w3": 1024,
    "model.layers.2.block_sparse_moe.experts.1.w1": 1024,
    "model.layers.2.block_sparse_moe.experts.1.w2": 1024,
    "model.layers.2.block_sparse_moe.experts.1.w3": 1024,
    "model.layers.2.block_sparse_moe.experts.2.w1": 1024,
    "model.layers.2.block_sparse_moe.experts.2.w2": 1024,
    "model.layers.2.block_sparse_moe.experts.2.w3": 1024,
    "model.layers.2.block_sparse_moe.experts.3.w1": 1024,
    "model.layers.2.block_sparse_moe.experts.3.w2": 1024,
    "model.layers.2.block_sparse_moe.experts.3.w3": 1024,
    "model.layers.2.block_sparse_moe.experts.4.w1": 1024,
    "model.layers.2.block_sparse_moe.experts.4.w2": 1024,
    "model.layers.2.block_sparse_moe.experts.4.w3": 1024,
    "model.layers.2.block_sparse_moe.experts.5.w1": 1024,
    "model.layers.2.block_sparse_moe.experts.5.w2": 1024,
    "model.layers.2.block_sparse_moe.experts.5.w3": 1024,
    "model.layers.2.block_sparse_moe.experts.6.w1": 1024,
    "model.layers.2.block_sparse_moe.experts.6.w2": 1024,
    "model.layers.2.block_sparse_moe.experts.6.w3": 1024,
    "model.layers.2.block_sparse_moe.experts.7.w1": 1024,
    "model.layers.2.block_sparse_moe.experts.7.w2": 1024,
    "model.layers.2.block_sparse_moe.experts.7.w3": 1024,
    "model.layers.3.block_sparse_moe.experts.0.w1": 1024,
    "model.layers.3.block_sparse_moe.experts.0.w2": 1024,
    "model.layers.3.block_sparse_moe.experts.0.w3": 1024,
    "model.layers.3.block_sparse_moe.experts.1.w1": 1024,
    "model.layers.3.block_sparse_moe.experts.1.w2": 1024,
    "model.layers.3.block_sparse_moe.experts.1.w3": 1024,
    "model.layers.3.block_sparse_moe.experts.2.w1": 1024,
    "model.layers.3.block_sparse_moe.experts.2.w2": 1024,
    "model.layers.3.block_sparse_moe.experts.2.w3": 1024,
    "model.layers.3.block_sparse_moe.experts.3.w1": 1024,
    "model.layers.3.block_sparse_moe.experts.3.w2": 1024,
    "model.layers.3.block_sparse_moe.experts.3.w3": 1024,
    "model.layers.3.block_sparse_moe.experts.4.w1": 1024,
    "model.layers.3.block_sparse_moe.experts.4.w2": 1024,
    "model.layers.3.block_sparse_moe.experts.4.w3": 1024,
    "model.layers.3.block_sparse_moe.experts.5.w1": 1024,
    "model.layers.3.block_sparse_moe.experts.5.w2": 1024,
    "model.layers.3.block_sparse_moe.experts.5.w3": 1024,
    "model.layers.3.block_sparse_moe.experts.6.w1": 1024,
    "model.layers.3.block_sparse_moe.experts.6.w2": 1024,
    "model.layers.3.block_sparse_moe.experts.6.w3": 1024,
    "model.layers.3.block_sparse_moe.experts.7.w1": 1024,
    "model.layers.3.block_sparse_moe.experts.7.w2": 1024,
    "model.layers.3.block_sparse_moe.experts.7.w3": 1024,
    "model.layers.4.block_sparse_moe.experts.0.w1": 1024,
    "model.layers.4.block_sparse_moe.experts.0.w2": 1024,
    "model.layers.4.block_sparse_moe.experts.0.w3": 1024,
    "model.layers.4.block_sparse_moe.experts.1.w1": 1024,
    "model.layers.4.block_sparse_moe.experts.1.w2": 1024,
    "model.layers.4.block_sparse_moe.experts.1.w3": 1024,
    "model.layers.4.block_sparse_moe.experts.2.w1": 1024,
    "model.layers.4.block_sparse_moe.experts.2.w2": 1024,
    "model.layers.4.block_sparse_moe.experts.2.w3": 1024,
    "model.layers.4.block_sparse_moe.experts.3.w1": 1024,
    "model.layers.4.block_sparse_moe.experts.3.w2": 1024,
    "model.layers.4.block_sparse_moe.experts.3.w3": 1024,
    "model.layers.4.block_sparse_moe.experts.4.w1": 1024,
    "model.layers.4.block_sparse_moe.experts.4.w2": 1024,
    "model.layers.4.block_sparse_moe.experts.4.w3": 1024,
    "model.layers.4.block_sparse_moe.experts.5.w1": 1024,
    "model.layers.4.block_sparse_moe.experts.5.w2": 1024,
    "model.layers.4.block_sparse_moe.experts.5.w3": 1024,
    "model.layers.4.block_sparse_moe.experts.6.w1": 1024,
    "model.layers.4.block_sparse_moe.experts.6.w2": 1024,
    "model.layers.4.block_sparse_moe.experts.6.w3": 1024,
    "model.layers.4.block_sparse_moe.experts.7.w1": 1024,
    "model.layers.4.block_sparse_moe.experts.7.w2": 1024,
    "model.layers.4.block_sparse_moe.experts.7.w3": 1024,
    "model.layers.5.block_sparse_moe.experts.0.w1": 1024,
    "model.layers.5.block_sparse_moe.experts.0.w2": 1024,
    "model.layers.5.block_sparse_moe.experts.0.w3": 1024,
    "model.layers.5.block_sparse_moe.experts.1.w1": 1024,
    "model.layers.5.block_sparse_moe.experts.1.w2": 1024,
    "model.layers.5.block_sparse_moe.experts.1.w3": 1024,
    "model.layers.5.block_sparse_moe.experts.2.w1": 1024,
    "model.layers.5.block_sparse_moe.experts.2.w2": 1024,
    "model.layers.5.block_sparse_moe.experts.2.w3": 1024,
    "model.layers.5.block_sparse_moe.experts.3.w1": 1024,
    "model.layers.5.block_sparse_moe.experts.3.w2": 1024,
    "model.layers.5.block_sparse_moe.experts.3.w3": 1024,
    "model.layers.5.block_sparse_moe.experts.4.w1": 1024,
    "model.layers.5.block_sparse_moe.experts.4.w2": 1024,
    "model.layers.5.block_sparse_moe.experts.4.w3": 1024,
    "model.layers.5.block_sparse_moe.experts.5.w1": 1024,
    "model.layers.5.block_sparse_moe.experts.5.w2": 1024,
    "model.layers.5.block_sparse_moe.experts.5.w3": 1024,
    "model.layers.5.block_sparse_moe.experts.6.w1": 1024,
    "model.layers.5.block_sparse_moe.experts.6.w2": 1024,
    "model.layers.5.block_sparse_moe.experts.6.w3": 1024,
    "model.layers.5.block_sparse_moe.experts.7.w1": 1024,
    "model.layers.5.block_sparse_moe.experts.7.w2": 1024,
    "model.layers.5.block_sparse_moe.experts.7.w3": 1024,
    "model.layers.6.block_sparse_moe.experts.0.w1": 1024,
    "model.layers.6.block_sparse_moe.experts.0.w2": 1024,
    "model.layers.6.block_sparse_moe.experts.0.w3": 1024,
    "model.layers.6.block_sparse_moe.experts.1.w1": 1024,
    "model.layers.6.block_sparse_moe.experts.1.w2": 1024,
    "model.layers.6.block_sparse_moe.experts.1.w3": 1024,
    "model.layers.6.block_sparse_moe.experts.2.w1": 1024,
    "model.layers.6.block_sparse_moe.experts.2.w2": 1024,
    "model.layers.6.block_sparse_moe.experts.2.w3": 1024,
    "model.layers.6.block_sparse_moe.experts.3.w1": 1024,
    "model.layers.6.block_sparse_moe.experts.3.w2": 1024,
    "model.layers.6.block_sparse_moe.experts.3.w3": 1024,
    "model.layers.6.block_sparse_moe.experts.4.w1": 1024,
    "model.layers.6.block_sparse_moe.experts.4.w2": 1024,
    "model.layers.6.block_sparse_moe.experts.4.w3": 1024,
    "model.layers.6.block_sparse_moe.experts.5.w1": 1024,
    "model.layers.6.block_sparse_moe.experts.5.w2": 1024,
    "model.layers.6.block_sparse_moe.experts.5.w3": 1024,
    "model.layers.6.block_sparse_moe.experts.6.w1": 1024,
    "model.layers.6.block_sparse_moe.experts.6.w2": 1024,
    "model.layers.6.block_sparse_moe.experts.6.w3": 1024,
    "model.layers.6.block_sparse_moe.experts.7.w1": 1024,
    "model.layers.6.block_sparse_moe.experts.7.w2": 1024,
    "model.layers.6.block_sparse_moe.experts.7.w3": 1024,
    "model.layers.7.block_sparse_moe.experts.0.w1": 1024,
    "model.layers.7.block_sparse_moe.experts.0.w2": 1024,
    "model.layers.7.block_sparse_moe.experts.0.w3": 1024,
    "model.layers.7.block_sparse_moe.experts.1.w1": 1024,
    "model.layers.7.block_sparse_moe.experts.1.w2": 1024,
    "model.layers.7.block_sparse_moe.experts.1.w3": 1024,
    "model.layers.7.block_sparse_moe.experts.2.w1": 1024,
    "model.layers.7.block_sparse_moe.experts.2.w2": 1024,
    "model.layers.7.block_sparse_moe.experts.2.w3": 1024,
    "model.layers.7.block_sparse_moe.experts.3.w1": 1024,
    "model.layers.7.block_sparse_moe.experts.3.w2": 1024,
    "model.layers.7.block_sparse_moe.experts.3.w3": 1024,
    "model.layers.7.block_sparse_moe.experts.4.w1": 1024,
    "model.layers.7.block_sparse_moe.experts.4.w2": 1024,
    "model.layers.7.block_sparse_moe.experts.4.w3": 1024,
    "model.layers.7.block_sparse_moe.experts.5.w1": 1024,
    "model.layers.7.block_sparse_moe.experts.5.w2": 1024,
    "model.layers.7.block_sparse_moe.experts.5.w3": 1024,
    "model.layers.7.block_sparse_moe.experts.6.w1": 1024,
    "model.layers.7.block_sparse_moe.experts.6.w2": 1024,
    "model.layers.7.block_sparse_moe.experts.6.w3": 1024,
    "model.layers.7.block_sparse_moe.experts.7.w1": 1024,
    "model.layers.7.block_sparse_moe.experts.7.w2": 1024,
    "model.layers.7.block_sparse_moe.experts.7.w3": 1024,
    "model.layers.8.block_sparse_moe.experts.0.w1": 1024,
    "model.layers.8.block_sparse_moe.experts.0.w2": 1024,
    "model.layers.8.block_sparse_moe.experts.0.w3": 1024,
    "model.layers.8.block_sparse_moe.experts.1.w1": 1024,
    "model.layers.8.block_sparse_moe.experts.1.w2": 1024,
    "model.layers.8.block_sparse_moe.experts.1.w3": 1024,
    "model.layers.8.block_sparse_moe.experts.2.w1": 1024,
    "model.layers.8.block_sparse_moe.experts.2.w2": 1024,
    "model.layers.8.block_sparse_moe.experts.2.w3": 1024,
    "model.layers.8.block_sparse_moe.experts.3.w1": 1024,
    "model.layers.8.block_sparse_moe.experts.3.w2": 1024,
    "model.layers.8.block_sparse_moe.experts.3.w3": 1024,
    "model.layers.8.block_sparse_moe.experts.4.w1": 1024,
    "model.layers.8.block_sparse_moe.experts.4.w2": 1024,
    "model.layers.8.block_sparse_moe.experts.4.w3": 1024,
    "model.layers.8.block_sparse_moe.experts.5.w1": 1024,
    "model.layers.8.block_sparse_moe.experts.5.w2": 1024,
    "model.layers.8.block_sparse_moe.experts.5.w3": 1024,
    "model.layers.8.block_sparse_moe.experts.6.w1": 1024,
    "model.layers.8.block_sparse_moe.experts.6.w2": 1024,
    "model.layers.8.block_sparse_moe.experts.6.w3": 1024,
    "model.layers.8.block_sparse_moe.experts.7.w1": 1024,
    "model.layers.8.block_sparse_moe.experts.7.w2": 1024,
    "model.layers.8.block_sparse_moe.experts.7.w3": 1024,
    "model.layers.9.block_sparse_moe.experts.0.w1": 1280,
    "model.layers.9.block_sparse_moe.experts.0.w2": 1536,
    "model.layers.9.block_sparse_moe.experts.0.w3": 1536,
    "model.layers.9.block_sparse_moe.experts.1.w1": 1536,
    "model.layers.9.block_sparse_moe.experts.1.w2": 1536,
    "model.layers.9.block_sparse_moe.experts.1.w3": 1536,
    "model.layers.9.block_sparse_moe.experts.2.w1": 1536,
    "model.layers.9.block_sparse_moe.experts.2.w2": 1536,
    "model.layers.9.block_sparse_moe.experts.2.w3": 1536,
    "model.layers.9.block_sparse_moe.experts.3.w1": 1536,
    "model.layers.9.block_sparse_moe.experts.3.w2": 1536,
    "model.layers.9.block_sparse_moe.experts.3.w3": 1536,
    "model.layers.9.block_sparse_moe.experts.4.w1": 1536,
    "model.layers.9.block_sparse_moe.experts.4.w2": 1536,
    "model.layers.9.block_sparse_moe.experts.4.w3": 1536,
    "model.layers.9.block_sparse_moe.experts.5.w1": 1536,
    "model.layers.9.block_sparse_moe.experts.5.w2": 1536,
    "model.layers.9.block_sparse_moe.experts.5.w3": 1536,
    "model.layers.9.block_sparse_moe.experts.6.w1": 1536,
    "model.layers.9.block_sparse_moe.experts.6.w2": 1536,
    "model.layers.9.block_sparse_moe.experts.6.w3": 1536,
    "model.layers.9.block_sparse_moe.experts.7.w1": 1536,
    "model.layers.9.block_sparse_moe.experts.7.w2": 1536,
    "model.layers.9.block_sparse_moe.experts.7.w3": 1536
  },
  "rms_norm_eps": 1e-05,
  "rope_theta": 1000000.0,
  "router_aux_loss_coef": 0.02,
  "router_jitter_noise": 0.0,
  "sliding_window": null,
  "tie_word_embeddings": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.45.2",
  "use_cache": true,
  "vocab_size": 32000
}