config.json

{
  "architectures": [
    "Qwen2BMForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "auto_map": {
    "AutoConfig": "hdong0/Qwen2.5-Math-1.5B-baseline-init--configuration_qwen2bm.Qwen2BMConfig",
    "AutoModelForCausalLM": "hdong0/Qwen2.5-Math-1.5B-baseline-init--modeling_qwen2bm.Qwen2BMForCausalLM"
  },
  "batch_mean": false,
  "batch_num_attention_heads": 4,
  "batch_num_key_value_heads": 4,
  "batch_partitions": 1,
  "hidden_act": "silu",
  "hidden_size": 1536,
  "initializer_range": 0.02,
  "intermediate_size": 8960,
  "is_baseline": true,
  "keys_to_ignore_at_inference": [
    "past_key_values"
  ],
  "max_position_embeddings": 4096,
  "max_window_layers": 21,
  "model_type": "qwen2bm",
  "num_attention_heads": 12,
  "num_hidden_layers": 28,
  "num_key_value_heads": 2,
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 10000,
  "sliding_window": 4096,
  "thin_batch_mix": false,
  "tie_word_embeddings": true,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.52.0.dev0",
  "use_cache": true,
  "use_sliding_window": false,
  "vocab_size": 151936
}