[Improvement] Support load_dotenv

Quickstart.md

@@ -4,14 +4,34 @@ Before running the evaluation script, you need to **configure** the VLMs and set
 
 After that, you can use a single script `run.py` to inference and evaluate multiple VLMs and benchmarks at a same time.
 
-## Step0. Installation
+## Step0. Installation & Setup essential keys
+
+**Installation. **
 
 ```bash
 git clone https://github.com/open-compass/VLMEvalKit.git
 cd VLMEvalKit
 pip install -e .
 ```
 
+**Setup Keys.**
+
+- To infer with API models (GPT-4v, Gemini-Pro-V, etc.) or use LLM APIs as the **judge or choice extractor**, you need to first setup API keys. You can place the required keys in `$VLMEvalKit/.env` or directly set them as the environment variable. If you choose to create a `.env` file, its content will look like:
+
+  ```bash
+  # The .env file, place it under $VLMEvalKit
+  # Alles-apin-token, for intra-org use only
+  ALLES=
+  # API Keys of Proprietary VLMs
+  DASHSCOPE_API_KEY=
+  GOOGLE_API_KEY=
+  OPENAI_API_KEY=
+  OPENAI_API_BASE=
+  STEPAI_API_KEY=
+  ```
+
+- Fill the blanks with your API keys (if necessary). Those API keys will be automatically loaded when doing the inference and evaluation.
+
 ## Step1. Configuration
 
 **VLM Configuration**: All VLMs are configured in `vlmeval/config.py`, for some VLMs, you need to configure the code root (MiniGPT-4, PandaGPT, etc.) or the model_weight root (LLaVA-v1-7B, etc.) before conducting the evaluation. During evaluation, you should use the model name specified in `supported_VLM` in `vlmeval/config.py` to select the VLM. For MiniGPT-4 and InstructBLIP, you also need to modify the config files in `vlmeval/vlm/misc` to configure LLM path and ckpt path.

requirements.txt

@@ -11,6 +11,7 @@ pandas>=1.5.3
 pillow
 portalocker
 pycocoevalcap
+python-dotenv
 requests
 rich
 seaborn

run.py

@@ -144,4 +144,5 @@ def main():
 
 
 if __name__ == '__main__':
+    load_env()
     main()

vlmeval/__init__.py

@@ -9,3 +9,5 @@
 from .utils import *
 from .vlm import *
 from .config import *
+
+load_env()

vlmeval/evaluate/misc.py

@@ -1,10 +1,12 @@
 import os
 from vlmeval.api import OpenAIWrapper, OpenAIWrapperInternal
+from vlmeval.smp import load_env
 
 INTERNAL = os.environ.get('INTERNAL', 0)
 
 
 def build_judge(version, **kwargs):
+    load_env()
     model_map = {
         'gpt-4-turbo': 'gpt-4-1106-preview',
         'gpt-4-0613': 'gpt-4-0613',

vlmeval/smp/misc.py

@@ -147,3 +147,24 @@ def run_command(cmd):
     if isinstance(cmd, str):
         cmd = cmd.split()
     return subprocess.check_output(cmd)
+
+def load_env():
+    try:
+        import vlmeval
+    except ImportError:
+        warnings.warn('VLMEval is not installed. Failed to import environment variables from .env file. ')
+        return
+    pth = osp.realpath(vlmeval.__path__[0])
+    pth = osp.join(pth, '../.env')
+    pth = osp.realpath(pth)
+    if not osp.exists(pth):
+        warnings.warn(f'Did not detect the .env file at {pth}, failed to load. ')
+        return
+
+    from dotenv import dotenv_values
+    values = dotenv_values(pth)
+    for k, v in values.items():
+        if v is not None and len(v):
+            os.environ[k] = v
+    print(f'API Keys successfully loaded from {pth}')
+    return