{
  "_name_or_path": "bigscience/bigscience-small-testing",
  "apply_residual_connection_post_layernorm": false,
  "architectures": [
    "BloomModel"
  ],
  "attention_dropout": 0.1,
  "attention_softmax_in_fp32": true,
  "bias_dropout_fusion": true,
  "bos_token_id": 0,
  "eos_token_id": 0,
  "hidden_dropout": 0.1,
  "initializer_range": 0.02,
  "layer_norm_epsilon": 1e-05,
  "masked_softmax_fusion": true,
  "n_embed": 64,
  "n_inner": null,
  "n_layer": 2,
  "num_attention_heads": 8,
  "pad_token_id": 0,
  "pretraining_tp": 2,
  "seq_length": 20,
  "skip_bias_add": true,
  "slow_but_exact": true,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.21.0.dev0",
  "use_cache": false,
  "vocab_size": 250880
}