gpt3-finnish-13B / config.json
Villekom's picture
Sharded model to decrease size
339a18b
{
"_name_or_path": "/users/villekom/scratch/models/gpt3-finnish-13B/",
"apply_residual_connection_post_layernorm": false,
"architectures": [
"BloomModel"
],
"attention_dropout": 0.0,
"attention_softmax_in_fp32": true,
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_dropout": 0.0,
"hidden_size": 5120,
"initializer_range": 0.02,
"layer_norm_epsilon": 1e-05,
"masked_softmax_fusion": true,
"model_type": "bloom",
"n_head": 40,
"n_layer": 40,
"pad_token_id": 3,
"pretraining_tp": 1,
"slow_but_exact": false,
"torch_dtype": "float32",
"transformers_version": "4.24.0",
"use_cache": true,
"vocab_size": 131072
}