add environement variables

app.py

@@ -15,25 +15,38 @@ from huggingface_hub import InferenceClient
 import openai
 from openai import OpenAI
 from globalvars import API_BASE, API_KEY, intention_prompt
+from dotenv import load_dotenv
 
 os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:30'
 os.environ['CUDA_LAUNCH_BLOCKING'] = '1'
 os.environ['CUDA_CACHE_DISABLE'] = '1'
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
+def load_env_variables():
+    # Load the .env file
+    load_dotenv()
+
+    # Retrieve the environment variables
+    hf_token = os.getenv('HF_TOKEN')
+    yi_token = os.getenv('YI_TOKEN')
+
+    return hf_token, yi_token
+
+# Example usage
+hf_token, yi_token = load_env_variables()
 ## add chroma vector store
 
 
 ## use instruct embeddings
 # Load the tokenizer and model
-tokenizer = AutoTokenizer.from_pretrained('nvidia/NV-Embed-v1', trust_remote_code=True)
-model = AutoModel.from_pretrained('nvidia/NV-Embed-v1', trust_remote_code=True).to(device)
+tokenizer = AutoTokenizer.from_pretrained('nvidia/NV-Embed-v1', token = hf_token , trust_remote_code=True)
+model = AutoModel.from_pretrained('nvidia/NV-Embed-v1' , token = hf_token ,  trust_remote_code=True).to(device)
 
 
 ## Make intention Mapper 
 
 intention_client = OpenAI(
-    api_key=API_KEY,
+    api_key=yi_token,
     base_url=API_BASE
 )
 intention_completion = intention_client.chat.completions.create(

requirements.txt

@@ -3,4 +3,5 @@ einops
 sentence-transformers
 torch==2.2.0
 transformers
-openai
+openai
+python-dotenv