Clean up

Pipes.py

@@ -1,7 +1,7 @@
 import os
 import logging
 from dotenv import load_dotenv
-from ezlocalai.LLM import LLM
+from ezlocalai.LLM import LLM, is_vision_model
 from ezlocalai.STT import STT
 from ezlocalai.CTTS import CTTS
 from pyngrok import ngrok
@@ -58,8 +58,10 @@ def __init__(self):
         else:
             logging.info(f"[LLM] {self.current_llm} model loading. Please wait...")
             self.llm = LLM(model=self.current_llm)
+            if is_vision_model(self.current_llm):
+                if self.vlm is None:
+                    self.vlm = self.llm
             logging.info(f"[LLM] {self.current_llm} model loaded successfully.")
-
         NGROK_TOKEN = os.environ.get("NGROK_TOKEN", "")
         if NGROK_TOKEN:
             ngrok.set_auth_token(NGROK_TOKEN)

cuda.Dockerfile

@@ -15,7 +15,7 @@ COPY cuda-requirements.txt .
 RUN python3 -m pip install --upgrade pip cmake scikit-build setuptools wheel --no-cache-dir && \
     CMAKE_ARGS="-DLLAMA_CUBLAS=on" FORCE_CMAKE=1 pip install llama-cpp-python==0.2.55 --no-cache-dir && \
     pip install --no-cache-dir -r cuda-requirements.txt
-RUN git clone https://github.com/deepseek-ai/DeepSeek-VL deepseek && \
+RUN git clone https://github.com/Josh-XT/DeepSeek-VL deepseek && \
     cd deepseek && \
     pip install --no-cache-dir -e . && \
     cd ..