config.json

{
  "architectures": [
    "HF_GPTForCausalLM"
  ],
  "bos_token_id": 1,
  "conv_size": 4,
  "eos_token_id": 2,
  "expand_k": 1.0,
  "expand_v": 1.0,
  "hidden_size": null,
  "mode": "fused_chunk",
  "model_type": "hf_gpt",
  "name": "SWA_GatedDeltaNet4_flanorm_1.3B",
  "norm_eps": 1e-05,
  "norm_first": false,
  "num_heads": 4,
  "pad_token_id": 0,
  "qk_activation": "silu",
  "qk_norm": "l2",
  "tie_word_embeddings": false,
  "torch_dtype": "float32",
  "transformers_version": "4.43.2",
  "use_beta": true,
  "use_cache": true,
  "use_gate": false,
  "use_output_norm": true,
  "use_short_conv": true,
  "vocab_size": 32000
}