update

Author: sbalandi

tools/llm_bench/task/pipeline_utils.py

@@ -7,6 +7,7 @@
 from llm_bench_utils.memory_monitor import MemMonitorWrapper
 from pathlib import Path
 from typing import Any
+from abc import ABC, abstractmethod
 
 
 def execution_time_in_sec(func):
@@ -20,7 +21,7 @@ def time_wrapper(self, args, **kwargs):
     return time_wrapper
 
 
-class CommonPipeline:
+class CommonPipeline(ABC):
     DEFAULT_OUTPUT_TOKEN_SIZE = 512
 
     def __init__(self, model, tokenizer, model_args: dict, model_path: Path, mem_consumption_meter: MemMonitorWrapper):
@@ -108,8 +109,9 @@ def print_generated(self, iter_num: int, generation_result: Any, prompt_idx: int
             generation_result (Any): Output of generation.
             prompt_idx (int): Number of the prompt being processed.
         """
-        pass
+        raise NotImplementedError
 
+    @abstractmethod
     def gen_iterate_data(
         self,
         input_token_size: int,
@@ -143,6 +145,7 @@ def gen_iterate_data(
         """
         return {}
 
+    @abstractmethod
     def postprocess_output_info(
         self,
         generation_result: Any,
@@ -184,6 +187,7 @@ def postprocess_output_info(
         """
         return {}, []
 
+    @abstractmethod
     def run(self, input_text: str, iter_num: int, prompt_index: int, proc_id: int, bench_hook: object | None) -> tuple[dict, list]:
         """Run pipeline.
 

tools/llm_bench/task/text_reranker.py

@@ -388,21 +388,19 @@ def launch(pipeline: CommonPipeline, iter_num: int, prompt_idx: int, iter_timest
 def get_texts_from_file(args: dict) -> list:
     texts_list = []
     if args["rerank_texts_file"] is not None and args["rerank_texts"] is not None:
-        log.warning("--texts and --texts_file are set, they are conflict options, it will be used texts from --texts_file")
+        raise RuntimeError("== --texts and --texts_file are set together, they define lists of texts both, please choose one of them ==")
 
     if args["rerank_texts_file"] is not None:
         for input_texts_file in args["rerank_texts_file"]:
-            if input_texts_file.endswith(".jsonl"):
-                if os.path.exists(input_texts_file):
-                    log.info(f"Read texts from {input_texts_file}")
-                    with open(input_texts_file, "r", encoding="utf-8") as f:
-                        for line in f:
-                            data = json.loads(line)
-                            texts_list.append(data["text"])
-                else:
-                    raise RuntimeError(f"== The texts file:{input_texts_file} does not exist ==")
-            else:
+            if not input_texts_file.endswith(".jsonl"):
                 raise RuntimeError(f"== The texts file:{input_texts_file} should be ended with .jsonl ==")
+            if not os.path.exists(input_texts_file):
+                raise RuntimeError(f"== The texts file:{input_texts_file} does not exist ==")
+            log.info(f"Read texts from {input_texts_file}")
+            with open(input_texts_file, "r", encoding="utf-8") as f:
+                for line in f:
+                    data = json.loads(line)
+                    texts_list.append(data["text"])
     else:
         if args["rerank_texts"] is not None:
             texts_list = args["rerank_texts"]