获取指定层的KV缓存（反量化后） Args: layer_idx: 层索引 Returns: Tuple[torch.Tensor, torch.Tensor]: (key_cache, value_cache)
(self, layer_idx)
| 206 | self.value_cache = [] |
| 207 | |
| 208 | def get_kvcache(self, layer_idx): |
| 209 | """ |
| 210 | 获取指定层的KV缓存（反量化后） |
| 211 | |
| 212 | Args: |
| 213 | layer_idx: 层索引 |
| 214 | |
| 215 | Returns: |
| 216 | Tuple[torch.Tensor, torch.Tensor]: (key_cache, value_cache) |
| 217 | """ |
| 218 | dequant_key = self._dequantize(self._quantized_key_cache[layer_idx]) |
| 219 | dequant_value = self._dequantize(self._quantized_value_cache[layer_idx]) |
| 220 | return dequant_key, dequant_value |
| 221 | |
| 222 | def update( |
| 223 | self, |
nothing calls this directly
no outgoing calls
no test coverage detected