[Feature] Add double order of subjective evaluation and removing duplicated response among two models (#692)

* add features

* add doc string

* add doc string

configs/datasets/subjective_cmp/subjective_corev2.py

@@ -36,7 +36,7 @@ for _name in subjective_all_sets:
     subjective_eval_cfg = dict(
         evaluator=dict(
             type=LMEvaluator,
-            random_order=True,
+            infer_order='random',
             prompt_template=dict(
                 type=PromptTemplate,
                 template=dict(round=[

configs/datasets/subjective_cmp/subjective_creation.py

@@ -34,7 +34,6 @@ for _name in subjective_all_sets:
     subjective_eval_cfg = dict(
         evaluator=dict(
             type=LMEvaluator,
-            random_order=True,
             prompt_template=dict(
                 type=PromptTemplate,
                 template=dict(round=[

docs/en/advanced_guides/subjective_evaluation.md

@@ -83,6 +83,10 @@ summarizer = dict(
 )
 ```
 
+In addition, you can also change the response order of the two models, please refer to `config/subjective_compare.py`,
+when `infer_order` is setting to `random`, the response will be random ordered,
+when `infer_order` is setting to `double`, the response of two models will be doubled in two ways.
+
 ### Single Model Scoring Configuration
 
 For `config/subjective_score.py`, it is mainly same with `config/subjective_compare.py`, and you just need to modify the eval mode to `singlescore`.

docs/zh_cn/advanced_guides/subjective_evaluation.md

@@ -83,6 +83,10 @@ summarizer = dict(
 )
 ```
 
+此外，在数据集的配置config中，还可以选择两回答比较时的回答顺序，请参考`config/subjective_compare.py`,
+当`infer_order`设置为`random`时，将对两模型的回复顺序进行随机打乱,
+当`infer_order`设置为`double`时，将把两模型的回复按两种先后顺序进行判断。
+
 ### 单回答打分配置
 
 对于单回答打分，更详细的config setting请参考 `config/subjective_score.py`，该config的大部分都与两回答比较的config相同，只需要修改评测模式即可，将评测模式设置为`singlescore`。

opencompass/openicl/icl_evaluator/lm_evaluator.py

@@ -1,8 +1,10 @@
+# flake8: noqa: E501
 import os.path as osp
 import random
 from typing import Dict, List, Optional
 
 import mmengine
+from datasets import Dataset
 from mmengine.config import ConfigDict
 
 from opencompass.openicl.icl_inferencer import GenInferencer
@@ -14,20 +16,40 @@ from opencompass.utils.text_postprocessors import first_number_postprocess
 from opencompass.utils.types import get_type_from_cfg
 
 
-def randomize_preds_and_record_references(predictions,
+def order_preds_and_record_references(predictions,
                                       references,
-                                          random_order,
+                                      infer_order,
                                       seed=2680):
+    """Order predictions based on args and recording regrading references.
+
+    Args:
+        predictions (List): List of multi model predictions.
+        references (List): List of reference based on each problem.
+        infer_order (str, optional): The mode of inference order.
+        seed (int, optional): Random seed.
+    """
     random.seed(seed)
     list_of_preds = [[] for _ in range(len(predictions))]
     for i in range(len(predictions[0]['model_preds'])):
         preds = [[pred['model_preds'][i], pred['model_name']]
                  for pred in predictions]
-        if random_order:
+        if infer_order == 'random':
             random.shuffle(preds)
         for j in range(len(preds)):
             list_of_preds[j].append(preds[j][0])
             references[i][f'answer{j+1}'] = preds[j][1]
+    if infer_order == 'double':
+        assert len(predictions) == 2
+        list_of_preds = [
+            a + b for a, b in zip(list_of_preds, reversed(list_of_preds))
+        ]
+        reversed_references = []
+        for item in references:
+            reversed_item = item.copy()
+            reversed_item['answer1'], reversed_item['answer2'] = reversed_item[
+                'answer2'], reversed_item['answer1']
+            reversed_references.append(reversed_item)
+        references += reversed_references
     return list_of_preds, references
 
 
@@ -52,10 +74,11 @@ class LMEvaluator:
         prompt_template: ConfigDict,
         judge_cfg: ConfigDict,
         output_path: str,
-        random_order: Optional[bool] = False,
+        infer_order: Optional[str] = 'random',
         dataset_cfg: Optional[ConfigDict] = None,
         postprocessor: ConfigDict = dict(type=first_number_postprocess)
     ) -> None:
+        assert infer_order in ['random', 'double']
         self.output_path = output_path
         out_dir, out_name = osp.split(output_path)
         if not out_dir:
@@ -74,20 +97,36 @@ class LMEvaluator:
         self.postprocessor = get_type_from_cfg(postprocessor)
         self.logger = get_logger()
         self.dataset_cfg = dataset_cfg
-        self.random_order = random_order
+        self.infer_order = infer_order
 
     def score(self, predictions, references: Optional[List] = None) -> Dict:
         if type(predictions) == list:
             """Apply to multi-model comparison."""
             references = [{} for _ in range(len(predictions[0]['model_preds']))
                           ] if references is None else references
-            predictions, references = randomize_preds_and_record_references(
+            predictions, references = order_preds_and_record_references(
-                predictions, references, self.random_order)
+                predictions, references, self.infer_order)
         elif type(predictions) == dict:
             """Apply to single-model scoring."""
             references = [{} for _ in range(len(predictions[0]['model_preds']))
                           ] if references is None else references
             predictions = [predictions['model_preds']]
+
+        # calculate dupicated predictions numbers
+        total_predictions_num = len(predictions[0])
+        dup_indices = []
+        for i in range(len(predictions[0])):
+            check = [sub[i] for sub in predictions]
+            if len(set(check)) == 1:
+                dup_indices.append(i)
+
+        if len(dup_indices) != 0:
+            # remove dupicated predictions
+            for index in sorted(dup_indices, reverse=True):
+                for sublist in predictions:
+                    del sublist[index]
+                del references[index]
+
         pred_dict = {}
         for i in range(len(predictions)):
             key = 'prediction' if i == 0 else f'prediction{i + 1}'
@@ -95,6 +134,25 @@ class LMEvaluator:
 
         if self.dataset_cfg:
             dataset = build_dataset_from_cfg(self.dataset_cfg)
+
+            if self.infer_order == 'double':
+                new_ds = {
+                    k: dataset.test[k] * 2
+                    for k in dataset.test.column_names
+                }
+                dataset.reader.dataset['test'] = Dataset.from_dict(new_ds)
+
+            if len(dup_indices) != 0:
+                remaining_indices = [
+                    idx for idx in range(len(dataset.test))
+                    if idx not in dup_indices
+                ]
+                dataset.reader.dataset['test'] = dataset.test.select(
+                    remaining_indices)
+                print(
+                    f'Among total {total_predictions_num} predictions, there are {len(dup_indices)} predictions totally same, which are removed!'
+                )
+
             for k, v in pred_dict.items():
                 dataset.reader.dataset['test'] = dataset.test.add_column(k, v)
                 dataset.reader.input_columns.append(k)