| { | |
| "vocab_size": 48000, | |
| "max_seq_len": 2048, | |
| "hidden_size": 768, | |
| "intermediate_size": 2304, | |
| "num_layers": 24, | |
| "num_heads": 12, | |
| "num_kv_heads": 4, | |
| "rope_theta": 10000.0, | |
| "rms_norm_eps": 1e-05, | |
| "qk_norm": true, | |
| "initializer_range": 0.02, | |
| "eos_token_id": 0, | |
| "pad_token_id": 1, | |
| "bos_token_id": 3, | |
| "tie_word_embeddings": true, | |
| "z_loss_coef": 0.0001, | |
| "attn_dropout": 0.0, | |
| "resid_dropout": 0.0 | |
| } |