[nfc] fix dim not defined and fix typo

applications/Chat/evaluate/gpt_evaluate.py

@@ -361,7 +361,7 @@ def get_gpt_evaluation_without_logprobs(prompt: Dict[str, Any],
     """
     Use chat models(gpt-3.5-turbo or gpt-4) to evaluate one model answer.
 
-    Temprature is set to 0 to make the model more deterministic.
+    Temperature is set to 0 to make the model more deterministic.
 
     Args:
         prompt: a dictionary including prompt template, CoT and metrics.
@@ -435,7 +435,7 @@ def get_gpt_evaluation_with_logprobs(prompt: Dict[str, Any],
     Use completion model(text-davinci-003) to evaluate one model answer.
     Only completion models can return log probabilities.
 
-    Temprature is set to 0 to make the model more deterministic.
+    Temperature is set to 0 to make the model more deterministic.
 
     Args:
         prompt: a dictionary including prompt template, CoT and metrics.
@@ -593,7 +593,7 @@ def calculate_scores_form_logprobs(logprobs: Dict[str, Any]) -> float:
 def calculate_scores_form_response(response: str, evaluation: Dict[str, Any]) -> int:
     """
     Calculate the score from the response returned by gpt-3.5-turbo or gpt-4.
-    Different from text-davinci-003, this fuction directly calculates the score according to the plain response returned by gpt-3.5-turbo or gpt-4.
+    Different from text-davinci-003, this function directly calculates the score according to the plain response returned by gpt-3.5-turbo or gpt-4.
     Although text-davinci-003 can return log probabilities, it costs ten times as much as gpt-3.5-turbo.
 
     Args:

applications/Chat/evaluate/unieval/evaluator.py

@@ -277,7 +277,7 @@ def evaluate(self, data, category):
         n_data = len(data)
         eval_scores = [{} for _ in range(n_data)]
 
-        # Calculate average sentence-level scores for facutal consistency
+        # Calculate average sentence-level scores for factual consistency
         src_list, output_list = [], []
         n_sents = []    # the number of sentences in the claim
         for i in range(n_data):
@@ -288,7 +288,7 @@ def evaluate(self, data, category):
                 src_list.append(source)
                 output_list.append(system_outputs[j])
         input_list = add_question(dimension=self.dim, output=output_list, src=src_list, task=self.task)
-        sent_score = self.scorer.score(input_list, self.task, category, dim)
+        sent_score = self.scorer.score(input_list, self.task, category, self.dim)
 
         # Get average score for each sample
         start_idx = 0

applications/Chat/evaluate/unieval/utils.py

@@ -37,7 +37,7 @@ def add_question(dimension, output, src=None, ref=None, context=None, task=None)
         src: source input for different NLG tasks. For example, source document for summarization
              and dialogue history for dialogue response generation.
         output: output text generated by the models
-        ref: human-annotataed groundtruth
+        ref: human-annotated groundtruth
         context: the context needed to evaluate several specific dimension. For example,
                  additional factual information when evaluating engagingness and groundedness in dialogues.
     """

applications/Chat/tests/test_data.py

@@ -33,7 +33,7 @@ def gather_and_equal(tensor: torch.Tensor) -> bool:
 
 
 def run_test_data(strategy):
-    EXPERINCE_BATCH_SIZE = 4
+    EXPERIENCE_BATCH_SIZE = 4
     SAMPLE_BATCH_SIZE = 2
 
     if strategy == 'ddp':
@@ -54,7 +54,7 @@ def run_test_data(strategy):
 
     # experience of all ranks should be the same
     for _ in range(2):
-        data = get_data(EXPERINCE_BATCH_SIZE)
+        data = get_data(EXPERIENCE_BATCH_SIZE)
         assert gather_and_equal(data['input_ids'])
         assert gather_and_equal(data['attention_mask'])
         experience = experience_maker.make_experience(**data,