[Mistral] Mistral-7B-v0.1 support (#1196)

Co-authored-by: timlacroix <t@mistral.ai>

vllm/engine/arg_utils.py

@@ -176,9 +176,9 @@ class EngineArgs:
                                    self.download_dir, self.load_format,
                                    self.dtype, self.seed, self.revision,
                                    self.max_model_len, self.quantization)
-        cache_config = CacheConfig(self.block_size,
-                                   self.gpu_memory_utilization,
-                                   self.swap_space)
+        cache_config = CacheConfig(
+            self.block_size, self.gpu_memory_utilization, self.swap_space,
+            getattr(model_config.hf_config, 'sliding_window', None))
         parallel_config = ParallelConfig(self.pipeline_parallel_size,
                                          self.tensor_parallel_size,
                                          self.worker_use_ray)

vllm/engine/llm_engine.py

@@ -86,6 +86,8 @@ class LLMEngine:
 
         self.model_config = model_config
         self.cache_config = cache_config
+        assert self.cache_config.sliding_window == getattr(
+            self.model_config.hf_config, "sliding_window", None)
         self.parallel_config = parallel_config
         self.scheduler_config = scheduler_config
         self.log_stats = log_stats