| from transformers.configuration_utils import PretrainedConfig | |
| from transformers.utils import logging | |
| logger = logging.get_logger(__name__) | |
| CM_PRETRAINED_CONFIG_ARCHIVE_MAP = {} | |
| class JiutianConfig(PretrainedConfig): | |
| model_type = "jiutian" | |
| keys_to_ignore_at_inference = ["past_key_values"] | |
| def __init__( | |
| self, | |
| vocab_size=152064, | |
| hidden_size=8192, | |
| intermediate_size=13312, | |
| num_hidden_layers=32, | |
| num_attention_heads=32, | |
| num_key_value_heads=8, | |
| hidden_act="silu", | |
| max_position_embeddings=8192, | |
| initializer_range=0.02, | |
| rms_norm_eps=1e-6, | |
| use_cache=True, | |
| pad_token_id=151645, | |
| bos_token_id=None, | |
| eos_token_id=151645, | |
| pretraining_tp=1, | |
| tie_word_embeddings=False, | |
| rope_theta=500000, | |
| rope_scaling=None, | |
| qkv_bias=True, | |
| attention_dropout=0.0, | |
| **kwargs, | |
| ): | |
| self.vocab_size = vocab_size | |
| self.max_position_embeddings = max_position_embeddings | |
| self.hidden_size = hidden_size | |
| self.intermediate_size = intermediate_size | |
| self.num_hidden_layers = num_hidden_layers | |
| self.num_attention_heads = num_attention_heads | |
| self.hidden_act = hidden_act | |
| self.initializer_range = initializer_range | |
| self.rms_norm_eps = rms_norm_eps | |
| self.pretraining_tp = pretraining_tp | |
| self.use_cache = use_cache | |
| self.rope_theta = rope_theta | |
| self.rope_scaling = None | |
| self.qkv_bias = qkv_bias | |
| self.attention_dropout = attention_dropout | |
| if num_key_value_heads is None: | |
| num_key_value_heads = num_attention_heads | |
| self.num_key_value_heads = num_key_value_heads | |
| super().__init__( | |
| pad_token_id=pad_token_id, | |
| bos_token_id=bos_token_id, | |
| eos_token_id=eos_token_id, | |
| tie_word_embeddings=tie_word_embeddings, | |
| **kwargs, | |
| ) | |