Hotfix bug from async refactor (#203)

memgpt/memory.py

@@ -382,7 +382,7 @@ class DummyArchivalMemoryWithFaiss(DummyArchivalMemory):
         if embedding is None:
             # Get the embedding
             embedding = async_get_embedding_with_backoff(memory_string, model=self.embedding_model)
-        return await self._insert(memory_string, embedding)
+        return self._insert(memory_string, embedding)
 
     def _search(self, query_embedding, query_string, count=None, start=None):
         """Simple embedding-based search (inefficient, no caching)"""
@@ -588,7 +588,7 @@ class DummyRecallMemoryWithEmbeddings(DummyRecallMemory):
         self.embedding_model = "text-embedding-ada-002"
         self.only_use_preloaded_embeddings = False
 
-    def _text_search(self, embedding_getter_func, query_string, count, start):
+    def text_search(self, query_string, count, start):
         # in the dummy version, run an (inefficient) case-insensitive match search
         message_pool = [d for d in self._message_logs if d["message"]["role"] not in ["system", "function"]]
 
@@ -603,11 +603,11 @@ class DummyRecallMemoryWithEmbeddings(DummyRecallMemory):
                     message_pool_filtered.append(d)
             elif message_str not in self.embeddings:
                 printd(f"recall_memory.text_search -- '{message_str}' was not in embedding dict, computing now")
-                self.embeddings[message_str] = embedding_getter_func(message_str, model=self.embedding_model)
+                self.embeddings[message_str] = get_embedding_with_backoff(message_str, model=self.embedding_model)
                 message_pool_filtered.append(d)
 
         # our wrapped version supports backoff/rate-limits
-        query_embedding = embedding_getter_func(query_string, model=self.embedding_model)
+        query_embedding = get_embedding_with_backoff(query_string, model=self.embedding_model)
         similarity_scores = [cosine_similarity(self.embeddings[d["message"]["content"]], query_embedding) for d in message_pool_filtered]
 
         # Sort the archive based on similarity scores
@@ -633,11 +633,8 @@ class DummyRecallMemoryWithEmbeddings(DummyRecallMemory):
         else:
             return matches, len(matches)
 
-    def text_search(self, query_string, count=None, start=None):
-        return self._text_search(get_embedding_with_backoff, query_string, count, start)
-
     async def a_text_search(self, query_string, count=None, start=None):
-        return await self._text_search(async_get_embedding_with_backoff, query_string, count, start)
+        return self.text_search(query_string, count, start)
 
 
 class LocalArchivalMemory(ArchivalMemory):