config bugfix

app.py

@@ -16,15 +16,6 @@ from backend.svg_processing import postprocess_svg
 
 logging.basicConfig(level=logging.DEBUG)
 
-with open("config.yaml") as stream:
-    try:
-        DEMO_CONFIG = yaml.safe_load(stream)
-        logging.debug(f"Config: {DEMO_CONFIG}")
-    except yaml.YAMLError as exc:
-        logging.error(f"Error loading config: {exc}")
-        raise exc
-
-
 
 EXAMPLES = [
     ("We're a nature-loving family with three kids, have some money left, and no plans "
@@ -94,7 +85,20 @@ CHATBOT_INSTRUCTIONS = (
 )
 
 # config
-client_kwargs, guide_kwargs = process_config(DEMO_CONFIG)
+with open("config.yaml") as stream:
+    try:
+        demo_config = yaml.safe_load(stream)
+        logging.debug(f"Config: {demo_config}")
+    except yaml.YAMLError as exc:
+        logging.error(f"Error loading config: {exc}")
+        gr.Error("Error loading config: {exc}")
+
+try:
+    client_kwargs, guide_kwargs = process_config(demo_config)
+except Exception as exc:
+    logging.error(f"Error processing config: {exc}")
+    gr.Error(f"Error processing config: {exc}")
+    
 logging.info(f"Reasoning guide expert model is {guide_kwargs['expert_model']}.")
 
 

backend/config.py

@@ -1,26 +1,5 @@
 import os
 
-# Default client
-INFERENCE_SERVER_URL = "https://api-inference.huggingface.co/models/{model_id}"
-MODEL_ID = "HuggingFaceH4/zephyr-7b-beta"
-CLIENT_MODEL_KWARGS = {
-    "max_tokens": 800,
-    "temperature": 0.6,
-}
-
-GUIDE_KWARGS = {
-    "expert_model": "HuggingFaceH4/zephyr-7b-beta",
-                  # "meta-llama/Meta-Llama-3.1-70B-Instruct",
-    "inference_server_url": "https://api-inference.huggingface.co/models/HuggingFaceH4/zephyr-7b-beta",
-                  # "https://api-inference.huggingface.co/models/meta-llama/Meta-Llama-3.1-70B-Instruct",
-    "llm_backend": "HFChat",
-    "classifier_kwargs": {
-        "model_id": "MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli",
-        "inference_server_url": "https://api-inference.huggingface.co/models/MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli",
-        "batch_size": 8,
-    },
-}
-
 
 def process_config(config):
     if "HF_TOKEN" not in os.environ:
@@ -37,8 +16,8 @@ def process_config(config):
             raise ValueError("config.yaml is missing client url.")
         client_kwargs["api_key"] = os.getenv("HF_TOKEN")
         client_kwargs["llm_backend"] = "HFChat"
-        client_kwargs["temperature"] = CLIENT_MODEL_KWARGS["temperature"]
-        client_kwargs["max_tokens"] = CLIENT_MODEL_KWARGS["max_tokens"]
+        client_kwargs["temperature"] = config["client_llm"].get("temperature",.6)
+        client_kwargs["max_tokens"] = config["client_llm"].get("max_tokens",800)
     else:
         raise ValueError("config.yaml is missing client_llm settings.")
     
@@ -67,7 +46,7 @@ def process_config(config):
         else:
             raise ValueError("config.yaml is missing classifier url.")
         if "batch_size" in config["classifier_llm"]:
-            guide_kwargs["classifier_kwargs"]["batch_size"] = config["classifier_llm"]["batch_size"]
+            guide_kwargs["classifier_kwargs"]["batch_size"] = int(config["classifier_llm"]["batch_size"])
         else:
             raise ValueError("config.yaml is missing classifier batch_size.")
         guide_kwargs["classifier_kwargs"]["api_key"] = os.getenv("HF_TOKEN")  # classifier api key

config.yaml

@@ -1,10 +1,12 @@
 client_llm:
   url: "https://api-inference.huggingface.co/models/HuggingFaceH4/zephyr-7b-beta"
   model_id: "HuggingFaceH4/zephyr-7b-beta"
+  max_tokens: 800
+  temperature: 0.6
 expert_llm:
   url: "https://api-inference.huggingface.co/models/HuggingFaceH4/zephyr-7b-beta"
   model_id: "HuggingFaceH4/zephyr-7b-beta"
 classifier_llm:
   model_id: "MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli"
   url: "https://api-inference.huggingface.co/models/MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli"
-  batch_size: 8,
+  batch_size: 8