fix: add sleep for too many requests

mteb/models/openai_models.py

@@ -19,7 +19,7 @@ def __init__(
         self,
         model_name: str,
         max_tokens: int,
-        tokenizer_name: str = "cl100k_base",  # since all models use this tokenizer now
+        tokenizer_name: str = "cl100k_base",  # since all models use this tokenizer now 
         embed_dim: int | None = None,
         **kwargs,
     ) -> None:
@@ -28,17 +28,23 @@ def __init__(
         """
         requires_package(self, "openai", "Openai text embedding")
         from openai import OpenAI
+        requires_package(self, "tiktoken", "Tiktoken package")
+        import tiktoken
 
         self._client = OpenAI()
         self._model_name = model_name
         self._embed_dim = embed_dim
         self._max_tokens = max_tokens
-        self._tokenizer_name = tokenizer_name
+        self._encoding = tiktoken.get_encoding(tokenizer_name)
+
+    def truncate_text_tokens(self, text):
+        """Truncate a string to have `max_tokens` according to the given encoding."""
+        truncated_sentence = self._encoding.encode(text)[:self._max_tokens]
+        return self._encoding.decode(truncated_sentence)
 
     def encode(self, sentences: list[str], **kwargs: Any) -> np.ndarray:
         requires_package(self, "openai", "Openai text embedding")
-        requires_package(self, "tiktoken", "Tiktoken package")
-        import tiktoken
+
         from openai import NotGiven
 
         if self._model_name == "text-embedding-ada-002" and self._embed_dim is not None:
@@ -48,11 +54,10 @@ def encode(self, sentences: list[str], **kwargs: Any) -> np.ndarray:
 
         trimmed_sentences = []
         for sentence in sentences:
-            encoding = tiktoken.get_encoding(self._tokenizer_name)
-            encoded_sentence = encoding.encode(sentence)
+            encoded_sentence = self._encoding.encode(sentence)
             if len(encoded_sentence) > self._max_tokens:
-                trimmed_sentence = encoding.decode(encoded_sentence[: self._max_tokens])
-                trimmed_sentences.append(trimmed_sentence)
+                truncated_sentence = self.truncate_text_tokens(sentence)
+                trimmed_sentences.append(truncated_sentence)
             else:
                 trimmed_sentences.append(sentence)
 
@@ -65,12 +70,34 @@ def encode(self, sentences: list[str], **kwargs: Any) -> np.ndarray:
         all_embeddings = []
 
         for sublist in sublists:
-            response = self._client.embeddings.create(
-                input=sublist,
-                model=self._model_name,
-                encoding_format="float",
-                dimensions=self._embed_dim or NotGiven(),
-            )
+            try:
+                response = self._client.embeddings.create(
+                    input=sublist,
+                    model=self._model_name,
+                    encoding_format="float",
+                    dimensions=self._embed_dim or NotGiven(),
+                )
+            except Exception as e:
+                # Sleep due to too many requests
+                logger.info("Sleeping for 10 seconds due to error", e)
+                import time
+                time.sleep(10)
+                try:
+                    response = self._client.embeddings.create(
+                        input=sublist,
+                        model=self._model_name,
+                        encoding_format="float",
+                        dimensions=self._embed_dim or NotGiven(),
+                    )
+                except Exception as e:
+                    logger.info("Sleeping for 60 seconds due to error", e)
+                    time.sleep(60)
+                    response = self._client.embeddings.create(
+                        input=sublist,
+                        model=self._model_name,
+                        encoding_format="float",
+                        dimensions=self._embed_dim or NotGiven(),
+                    )
             all_embeddings.extend(self._to_numpy(response))
 
         return np.array(all_embeddings)
@@ -126,7 +153,12 @@ def _to_numpy(self, embedding_response) -> np.ndarray:
     revision="2",
     release_date="2022-12-15",
     languages=None,  # supported languages not specified
-    loader=partial(OpenAIWrapper, model_name="text-embedding-ada-002"),
+    loader=partial(
+        OpenAIWrapper,
+        model_name="text-embedding-ada-002",
+        tokenizer_name="cl100k_base",
+        max_tokens=8192,
+    ),
     reference="https://openai.com/index/new-and-improved-embedding-model/",
     max_tokens=8191,
     embed_dim=1536,