configuration_modern_llm.py

from transformers import PretrainedConfig

QWEN_TOKEN_ID = 151643

AUTO_MAP = {
    "AutoConfig": "configuration_modern_llm.ModernLLMConfig",
    "AutoModelForCausalLM": "modeling_modern_llm.ModernLLMForCausalLM",
}


class ModernLLMConfig(PretrainedConfig):
    model_type = "modern_llm"

    def __init__(self, vocab_size=151936, hidden_size=768, intermediate_size=2048,
                 num_hidden_layers=12, num_attention_heads=12, num_key_value_heads=4,
                 max_position_embeddings=2048, rms_norm_eps=1e-6, rope_theta=1000000.0,
                 pad_token_id=QWEN_TOKEN_ID, bos_token_id=QWEN_TOKEN_ID, eos_token_id=QWEN_TOKEN_ID, **kwargs):
        self.vocab_size = vocab_size
        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.num_key_value_heads = num_key_value_heads
        self.max_position_embeddings = max_position_embeddings
        self.rms_norm_eps = rms_norm_eps
        self.rope_theta = rope_theta
        self.auto_map = dict(AUTO_MAP)
        super().__init__(pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)