modeling_dummy.py

import torch
from torch import nn
from transformers import PreTrainedModel, PretrainedConfig

class DummyConfig(PretrainedConfig):
    model_type = "dummy"
    
    def __init__(
        self,
        vocab_size=32000,
        hidden_size=32,
        intermediate_size=64,
        num_hidden_layers=1,
        num_attention_heads=1,
        max_position_embeddings=2048,
        pad_token_id=0,
        bos_token_id=1,
        eos_token_id=2,
        **kwargs
    ):
        super().__init__(
            pad_token_id=pad_token_id,
            bos_token_id=bos_token_id,
            eos_token_id=eos_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.max_position_embeddings = max_position_embeddings

class DummyForCausalLM(PreTrainedModel):
    config_class = DummyConfig
    _keys_to_ignore_on_load_missing = ["lm_head.weight"]
    
    def __init__(self, config):
        super().__init__(config)
        self.config = config
        self.embed = nn.Embedding(config.vocab_size, config.hidden_size)
        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
        
        # 고정 응답용 토큰
        self.fixed_response = "이것은 더미 모델의 고정 응답입니다. vLLM 서빙 테스트용으로 만들어졌습니다."
        
        # 가중치 초기화
        self.post_init()
        
    def get_input_embeddings(self):
        return self.embed
        
    def set_input_embeddings(self, value):
        self.embed = value
        
    def get_output_embeddings(self):
        return self.lm_head
    
    def forward(self, input_ids=None, attention_mask=None, **kwargs):
        batch_size = input_ids.shape[0] if input_ids is not None else 1
        seq_len = input_ids.shape[1] if input_ids is not None else 1
        
        # 매우 간단한 임베딩
        dummy_hidden = torch.zeros((batch_size, seq_len, self.config.hidden_size), 
                                    dtype=torch.float32, 
                                    device=input_ids.device if input_ids is not None else "cpu")
        
        # 임의의 로짓 값 생성
        logits = self.lm_head(dummy_hidden)
        
        # 항상 고정된 응답으로 예측되도록
        return {"logits": logits}
    
    def prepare_inputs_for_generation(self, input_ids, past_key_values=None, **kwargs):
        return {
            "input_ids": input_ids,
            "past_key_values": past_key_values
        }
    
    def _reorder_cache(self, past_key_values, beam_idx):
        return past_key_values