CerebrasGPT-1.3B_SP/params_train_to_hf.json

{
    "attn_pdrop": 0.0,
    "scale_attn_weights": true,
    "resid_pdrop": 0.0,
    "n_inner": 8192,
    "n_embd": 2048,
    "layer_norm_epsilon": 1e-05,
    "n_positions": 2048,
    "activation_function": "gelu",
    "n_head": 16,
    "n_layer": 24,
    "tie_word_embeddings": true,
    "vocab_size": 50257,
    "embd_pdrop": 0.0
}