Enhance autogptq backend to support VL models (#1860)

* Enhance autogptq backend to support VL models

* update dependencies for autogptq

* remove redundant auto-gptq dependency

* Convert base64 to image_url for Qwen-VL model

* implemented model inference for qwen-vl

* remove user prompt from generated answer

* fixed write image error

---------

Co-authored-by: Binghua Wu <bingwu@estee.com>

backend/python/common-env/transformers/transformers.yml

@@ -24,9 +24,10 @@ dependencies:
   - xz=5.4.2=h5eee18b_0
   - zlib=1.2.13=h5eee18b_0
   - pip:
-      - accelerate==0.23.0
+      - accelerate==0.27.0
       - aiohttp==3.8.5
       - aiosignal==1.3.1
+      - auto-gptq==0.7.1
       - async-timeout==4.0.3
       - attrs==23.1.0
       - bark==0.1.5
@@ -56,6 +57,7 @@ dependencies:
       - multiprocess==0.70.15
       - networkx
       - numpy==1.26.0
+      - optimum==1.17.1
       - packaging==23.2
       - pandas
       - peft==0.5.0
@@ -74,13 +76,13 @@ dependencies:
       - six==1.16.0
       - sympy==1.12
       - tokenizers
-      - torch==2.1.2
+      - torch==2.2.1
+      - torchvision==0.17.1
       - torchaudio==2.1.2
       - tqdm==4.66.1
       - triton==2.1.0
       - typing-extensions==4.8.0
       - tzdata==2023.3
-      - auto-gptq==0.6.0
       - urllib3==1.26.17
       - xxhash==3.4.1
       - yarl==1.9.2
@@ -103,5 +105,6 @@ dependencies:
       - vocos
       - vllm==0.3.2
       - transformers>=4.38.2  # Updated Version
+      - transformers_stream_generator==0.0.5
       - xformers==0.0.23.post1  
 prefix: /opt/conda/envs/transformers