config.json

{
  "architectures": [
    "CuriousForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "attn_logit_softcapping": 50.0,
  "bos_token_id": 2,
  "cache_implementation": "hybrid",
  "curious_architecture_type": "instruction_tuned_transformer",
  "curious_base_model": "adam-2b-original",
  "curious_capabilities": [
    "Natural conversations",
    "Question answering",
    "Creative writing",
    "Code assistance",
    "Mathematical reasoning",
    "Instruction following"
  ],
  "curious_features": [
    "Enhanced 2B parameter architecture",
    "Instruction-tuned for better conversations",
    "Advanced attention mechanisms",
    "Optimized activation functions",
    "Sliding window attention",
    "Logit softcapping for stability"
  ],
  "curious_instruction_tuned": true,
  "curious_model_size": "2B",
  "curious_training": "Instruction tuned on conversational data",
  "curious_version": "2.0",
  "dtype": "float32",
  "eos_token_id": [
    1,
    107
  ],
  "final_logit_softcapping": 30.0,
  "head_dim": 256,
  "hidden_act": "gelu_pytorch_tanh",
  "hidden_activation": "gelu_pytorch_tanh",
  "hidden_size": 2304,
  "initializer_range": 0.02,
  "intermediate_size": 9216,
  "layer_types": [
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention",
    "sliding_attention",
    "full_attention"
  ],
  "max_position_embeddings": 8192,
  "model_type": "curious_text",
  "num_attention_heads": 8,
  "num_hidden_layers": 26,
  "num_key_value_heads": 4,
  "pad_token_id": 0,
  "query_pre_attn_scalar": 256,
  "rms_norm_eps": 1e-06,
  "rope_theta": 10000.0,
  "sliding_window": 4096,
  "transformers_version": "4.56.2",
  "use_bidirectional_attention": null,
  "use_cache": true,
  "vocab_size": 256000
}