Fix transformers dependencies for slim. (#3934)

### What problem does this PR solve?


### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)

rag/llm/cv_model.py

@@ -25,7 +25,6 @@
 from io import BytesIO
 import json
 import requests
-from transformers import GenerationConfig
 
 from rag.nlp import is_english
 from api.utils import get_uuid
@@ -510,6 +509,7 @@ def describe(self, image, max_tokens=2048):
         return res.text,res.usage_metadata.total_token_count
 
     def chat(self, system, history, gen_conf, image=""):
+        from transformers import GenerationConfig
         if system:
             history[-1]["content"] = system + history[-1]["content"] + "user query: " + history[-1]["content"]
         try:
@@ -533,6 +533,7 @@ def chat(self, system, history, gen_conf, image=""):
             return "**ERROR**: " + str(e), 0
 
     def chat_streamly(self, system, history, gen_conf, image=""):
+        from transformers import GenerationConfig
         if system:
             history[-1]["content"] = system + history[-1]["content"] + "user query: " + history[-1]["content"]