ablation_initialization/mqa_normal_initialization/config.json

{
  "architectures": [
    "GPT"
  ],
  "bias": false,
  "block_size": 2048,
  "dropout": 0.0,
  "group_size": 24,
  "head_dim": 128,
  "intermediate_size": 10152,
  "model_type": "gpt2",
  "n_embd": 3072,
  "n_head": 24,
  "n_layer": 24,
  "torch_dtype": "float32",
  "transformers_version": "4.51.3",
  "vocab_size": 50304
}