switch to OpenAIServerModel

app.py

@@ -12,8 +12,9 @@ from dotenv import load_dotenv
 # from huggingface_hub import login
 from smolagents import (
     CodeAgent,
-    HfApiModel,
-    LiteLLMModel,
+    # HfApiModel,
+    # LiteLLMModel,
+    OpenAIServerModel,
     Tool,
     # GoogleSearchTool,
     DuckDuckGoSearchTool,
@@ -96,7 +97,8 @@ _ = "" if os.getenv("OPENAI_API_KEY") is None else os.getenv("OPENAI_API_KEY")[:
 
 if os.getenv("MODEL_ID") and os.getenv("OPENAI_API_BASE"):
     logger.debug(f"""using LiteLLMModel: {model_id=}, {os.getenv("OPENAI_API_BASE")=}, os.getenv("OPENAI_API_BASE")={_}""")
-    model = LiteLLMModel(
+    # model = LiteLLMModel(
+    model = OpenAIServerModel(
         # "gpt-4o",
         # os.getenv("MODEL_ID", "gpt-4o-mini"),
         model_id,
@@ -286,13 +288,6 @@ class GradioUI:
                     ):
                         file_uploads_log = gr.State([])
                         with gr.Sidebar():
-                            gr.Markdown("""# open Deep Research - free the AI agents!
-
-                OpenAI just (February 2, 2025) published [Deep Research](https://openai.com/index/introducing-deep-research/), an amazing assistant that can perform deep searches on the web to answer user questions.
-
-                However, their agent has a huge downside: it's not open. So we've started a 24-hour rush to replicate and open-source it. Our (Huggingface's) resulting [open-Deep-Research agent](https://github.com/huggingface/smolagents/tree/main/examples/open_deep_research) took the #1 rank of any open submission on the GAIA leaderboard! ✨
-
-                You can try a simplified version here that uses `Qwen-Coder-32B` (via smolagnet.HfApiModel) instead of `o1`. Modified: if you set MODEL_ID, OPENAI_API_BASE and OPENAI_API_KEY in the .env or env vars (in hf space these can be set in settings, .env will override env vars), the correspoding model will be used. N.B. if you see errors, it might be because whatever quota is exceeded, clone/duplicate this space and plug in your own resources and run your own deep-research.<br><br>""")
                             with gr.Group():
                                 gr.Markdown("**Your request**", container=True)
                                 text_input = gr.Textbox(
@@ -326,6 +321,15 @@ class GradioUI:
                         <a target="_blank" href="https://github.com/huggingface/smolagents"><b>huggingface/smolagents</b></a>
                         </div>""")
 
+                            # -----
+                            gr.Markdown("""# open Deep Research - free the AI agents!
+
+                OpenAI just (February 2, 2025) published [Deep Research](https://openai.com/index/introducing-deep-research/), an amazing assistant that can perform deep searches on the web to answer user questions.
+
+                However, their agent has a huge downside: it's not open. So we've started a 24-hour rush to replicate and open-source it. Our (Huggingface's) resulting [open-Deep-Research agent](https://github.com/huggingface/smolagents/tree/main/examples/open_deep_research) took the #1 rank of any open submission on the GAIA leaderboard! ✨
+
+                You can try a simplified version here that uses `Qwen-Coder-32B` (via smolagnet.HfApiModel) instead of `o1`. Modified: if you set MODEL_ID, OPENAI_API_BASE and OPENAI_API_KEY in the .env or env vars (in hf space these can be set in settings, .env will override env vars), the correspoding model will be used. N.B. if you see errors, it might be because whatever quota is exceeded, clone/duplicate this space and plug in your own resources and run your own deep-research.<br><br>""")
+
                         # Add session state to store session-specific data
                         session_state = gr.State(
                             {}
@@ -477,5 +481,8 @@ class GradioUI:
 
         demo.launch(debug=True, **kwargs)
 
-
-GradioUI().launch()
+# can this fix ctrl-c no response? no
+try:
+    GradioUI().launch()
+except KeyboardInterrupt:
+    ...

requirements.txt

@@ -1,4 +1,4 @@
-smolagents[litellm]==1.14.0
+smolagents
 anthropic>=0.37.1
 beautifulsoup4>=4.12.3
 datasets>=2.21.0
@@ -38,7 +38,7 @@ python-pptx
 torch
 xlrd
 SpeechRecognition
-litellm>=1.67.2
+# litellm>=1.67.2
 
 duckduckgo-search
 loguru