config.json

{
  "architectures": [
    "PatramForCausalLM"
  ],
  "attention_layer_norm": true,
  "auto_map": {
    "AutoConfig": "config_patram.PatramConfig",
    "AutoModelForCausalLM": "modeling_patram.PatramForCausalLM"
  },
  "clip_qkv": null,
  "embedding_size": 100352,
  "hidden_size": 4096,
  "initializer_range": 0.02,
  "intermediate_size": 22016,
  "layer_norm_eps": 1e-06,
  "layer_norm_type": "rms",
  "max_position_embeddings": 4096,
  "model_type": "patram",
  "norm_after": true,
  "num_attention_heads": 32,
  "num_hidden_layers": 32,
  "num_key_value_heads": null,
  "pad_token_id": 100277,
  "qkv_bias": false,
  "rope_theta": 500000.0,
  "tie_word_embeddings": false,
  "torch_dtype": "float32",
  "transformers_version": "4.52.3",
  "use_cache": false,
  "use_position_ids": true,
  "vocab_size": 100278,
  "weight_tying": false
}