| { | |
| "model_type": "gslm-unit-lm", | |
| "architectures": ["UnitLanguageModel"], | |
| "vocab_size": 205, | |
| "dict_file": "dict.txt", | |
| "pad_token_id": 1, | |
| "bos_token_id": 0, | |
| "eos_token_id": 2, | |
| "unk_token_id": 3, | |
| "n_layer": 12, | |
| "n_head": 12, | |
| "n_embd": 768, | |
| "max_position_embeddings": 4096, | |
| "dropout": 0.1, | |
| "attn_impl": "separate_qkv", | |
| "norm_type": "layernorm", | |
| "pos_embed": "sinusoidal", | |
| "bias": false, | |
| "tie_word_embeddings": true, | |
| "generation_prefix_size": -1, | |
| "generation_temperature": 0.7, | |
| "generation_sampling": true, | |
| "generation_beam": 1, | |
| "generation_max_len_a": 0.0, | |
| "generation_max_len_b": 1000, | |
| "generation_top_k": 0, | |
| "generation_top_p": 0.0 | |
| } | |