-- File downloading logic is added.
Browse files- app.py +31 -4
- langchain_agent.py +17 -16
app.py
CHANGED
|
@@ -1,3 +1,4 @@
|
|
|
|
|
| 1 |
import os
|
| 2 |
import gradio as gr
|
| 3 |
import requests
|
|
@@ -32,6 +33,7 @@ async def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 32 |
api_url = DEFAULT_API_URL
|
| 33 |
questions_url = f"{api_url}/questions"
|
| 34 |
submit_url = f"{api_url}/submit"
|
|
|
|
| 35 |
|
| 36 |
# 1. Instantiate Agent ( modify this part to create your agent)
|
| 37 |
try:
|
|
@@ -71,9 +73,31 @@ async def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 71 |
for item in questions_data[:]:
|
| 72 |
task_id = item.get("task_id")
|
| 73 |
question_text = item.get("question")
|
|
|
|
|
|
|
| 74 |
if not task_id or question_text is None:
|
| 75 |
print(f"Skipping item with missing task_id or question: {item}")
|
| 76 |
continue
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 77 |
try:
|
| 78 |
submitted_answer = await agent(question_text)
|
| 79 |
answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
|
|
@@ -82,6 +106,9 @@ async def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 82 |
print(f"Error running agent on task {task_id}: {e}")
|
| 83 |
results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
|
| 84 |
|
|
|
|
|
|
|
|
|
|
| 85 |
if not answers_payload:
|
| 86 |
print("Agent did not produce any answers to submit.")
|
| 87 |
return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
|
|
@@ -108,7 +135,7 @@ async def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 108 |
results_df = pd.DataFrame(results_log)
|
| 109 |
return final_status, results_df
|
| 110 |
except requests.exceptions.HTTPError as e:
|
| 111 |
-
error_detail = f"Server responded with status {e.response.status_code}."
|
| 112 |
try:
|
| 113 |
error_json = e.response.json()
|
| 114 |
error_detail += f" Detail: {error_json.get('detail', e.response.text)}"
|
|
@@ -139,7 +166,7 @@ async def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 139 |
with gr.Blocks() as demo:
|
| 140 |
gr.Markdown("# Basic Agent Evaluation Runner")
|
| 141 |
gr.Markdown(
|
| 142 |
-
|
| 143 |
**Instructions:**
|
| 144 |
|
| 145 |
1. Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
|
|
@@ -150,7 +177,7 @@ with gr.Blocks() as demo:
|
|
| 150 |
**Disclaimers:**
|
| 151 |
Once clicking on the "submit button, it can take quite some time ( this is the time for the agent to go through all the questions).
|
| 152 |
This space provides a basic setup and is intentionally sub-optimal to encourage you to develop your own, more robust solution. For instance for the delay process of the submit button, a solution could be to cache the answers and submit in a seperate action or even to answer the questions in async.
|
| 153 |
-
|
| 154 |
)
|
| 155 |
|
| 156 |
gr.LoginButton()
|
|
@@ -188,4 +215,4 @@ if __name__ == "__main__":
|
|
| 188 |
print("-"*(60 + len(" App Starting ")) + "\n")
|
| 189 |
|
| 190 |
print("Launching Gradio Interface for Basic Agent Evaluation...")
|
| 191 |
-
demo.launch(debug=True, share=False, server_name="0.0.0.0")
|
|
|
|
| 1 |
+
import asyncio
|
| 2 |
import os
|
| 3 |
import gradio as gr
|
| 4 |
import requests
|
|
|
|
| 33 |
api_url = DEFAULT_API_URL
|
| 34 |
questions_url = f"{api_url}/questions"
|
| 35 |
submit_url = f"{api_url}/submit"
|
| 36 |
+
file_url = f"{api_url}/files"
|
| 37 |
|
| 38 |
# 1. Instantiate Agent ( modify this part to create your agent)
|
| 39 |
try:
|
|
|
|
| 73 |
for item in questions_data[:]:
|
| 74 |
task_id = item.get("task_id")
|
| 75 |
question_text = item.get("question")
|
| 76 |
+
file_name = item.get("file_name")
|
| 77 |
+
|
| 78 |
if not task_id or question_text is None:
|
| 79 |
print(f"Skipping item with missing task_id or question: {item}")
|
| 80 |
continue
|
| 81 |
+
if file_name:
|
| 82 |
+
if not os.path.exists("resource"):
|
| 83 |
+
os.makedirs("resource")
|
| 84 |
+
try:
|
| 85 |
+
download_url = f"{file_url}/{task_id}"
|
| 86 |
+
response = requests.get(download_url)
|
| 87 |
+
response.raise_for_status()
|
| 88 |
+
file_path = os.path.join("resource", file_name)
|
| 89 |
+
with open(file_path, "wb") as f:
|
| 90 |
+
f.write(response.content)
|
| 91 |
+
print(f"Successfully downloaded {file_name} to {file_path}")
|
| 92 |
+
# Read the file content and pass it to the agent
|
| 93 |
+
with open(file_path, "r", encoding="utf-8") as f:
|
| 94 |
+
file_content = f.read()
|
| 95 |
+
question_text = f"{question_text}\n\nFile content:\n{file_content}"
|
| 96 |
+
except requests.exceptions.RequestException as e:
|
| 97 |
+
print(f"Error downloading file {file_name}: {e}")
|
| 98 |
+
except IOError as e:
|
| 99 |
+
print(f"Error reading file {file_name}: {e}")
|
| 100 |
+
|
| 101 |
try:
|
| 102 |
submitted_answer = await agent(question_text)
|
| 103 |
answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
|
|
|
|
| 106 |
print(f"Error running agent on task {task_id}: {e}")
|
| 107 |
results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
|
| 108 |
|
| 109 |
+
# Wait for a short moment to avoid overwhelming the server (optional)
|
| 110 |
+
await asyncio.sleep(1 * 60)
|
| 111 |
+
|
| 112 |
if not answers_payload:
|
| 113 |
print("Agent did not produce any answers to submit.")
|
| 114 |
return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
|
|
|
|
| 135 |
results_df = pd.DataFrame(results_log)
|
| 136 |
return final_status, results_df
|
| 137 |
except requests.exceptions.HTTPError as e:
|
| 138 |
+
error_detail = f"Server responded with status {e.response.status_code}. "
|
| 139 |
try:
|
| 140 |
error_json = e.response.json()
|
| 141 |
error_detail += f" Detail: {error_json.get('detail', e.response.text)}"
|
|
|
|
| 166 |
with gr.Blocks() as demo:
|
| 167 |
gr.Markdown("# Basic Agent Evaluation Runner")
|
| 168 |
gr.Markdown(
|
| 169 |
+
'''
|
| 170 |
**Instructions:**
|
| 171 |
|
| 172 |
1. Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
|
|
|
|
| 177 |
**Disclaimers:**
|
| 178 |
Once clicking on the "submit button, it can take quite some time ( this is the time for the agent to go through all the questions).
|
| 179 |
This space provides a basic setup and is intentionally sub-optimal to encourage you to develop your own, more robust solution. For instance for the delay process of the submit button, a solution could be to cache the answers and submit in a seperate action or even to answer the questions in async.
|
| 180 |
+
'''
|
| 181 |
)
|
| 182 |
|
| 183 |
gr.LoginButton()
|
|
|
|
| 215 |
print("-"*(60 + len(" App Starting ")) + "\n")
|
| 216 |
|
| 217 |
print("Launching Gradio Interface for Basic Agent Evaluation...")
|
| 218 |
+
demo.launch(debug=True, share=False, server_name="0.0.0.0")
|
langchain_agent.py
CHANGED
|
@@ -71,27 +71,28 @@ class LangChainAgent:
|
|
| 71 |
tools = await client.get_tools()
|
| 72 |
print(tools)
|
| 73 |
|
| 74 |
-
|
| 75 |
-
|
| 76 |
|
| 77 |
-
|
| 78 |
|
| 79 |
|
| 80 |
-
# model_name = "deepseek-ai/deepseek-v3.1"
|
| 81 |
-
# model_name = "deepseek-ai/deepseek-v3.1-terminus"
|
| 82 |
-
# model_name = "minimaxai/minimax-m2"
|
| 83 |
-
# model_name = "mistralai/mistral-nemotron"
|
| 84 |
-
# model_name = "qwen/qwen3-next-80b-a3b-instruct"
|
| 85 |
-
model_name = "qwen/qwen3-next-80b-a3b-thinking"
|
| 86 |
-
# model_name = "moonshotai/kimi-k2-instruct-0905"
|
| 87 |
-
#
|
|
|
|
| 88 |
|
| 89 |
|
| 90 |
-
model = ChatOpenAI(
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
)
|
| 95 |
|
| 96 |
agent = create_agent(model, tools)
|
| 97 |
|
|
|
|
| 71 |
tools = await client.get_tools()
|
| 72 |
print(tools)
|
| 73 |
|
| 74 |
+
model_name = "gemini-2.0-flash"
|
| 75 |
+
model_provider = "google_genai" #google_genai
|
| 76 |
|
| 77 |
+
model = init_chat_model(model_name, model_provider=model_provider)
|
| 78 |
|
| 79 |
|
| 80 |
+
# # model_name = "deepseek-ai/deepseek-v3.1"
|
| 81 |
+
# # model_name = "deepseek-ai/deepseek-v3.1-terminus"
|
| 82 |
+
# # model_name = "minimaxai/minimax-m2"
|
| 83 |
+
# # model_name = "mistralai/mistral-nemotron"
|
| 84 |
+
# # model_name = "qwen/qwen3-next-80b-a3b-instruct"
|
| 85 |
+
# model_name = "qwen/qwen3-next-80b-a3b-thinking"
|
| 86 |
+
# # model_name = "moonshotai/kimi-k2-instruct-0905"
|
| 87 |
+
# model_name = "nvidia/llama-3.3-nemotron-super-49b-v1.5"
|
| 88 |
+
# # model_provider = "nvidia"
|
| 89 |
|
| 90 |
|
| 91 |
+
# model = ChatOpenAI(
|
| 92 |
+
# model=model_name,
|
| 93 |
+
# openai_api_key=os.getenv("NVIDIA_API_KEY"),
|
| 94 |
+
# openai_api_base="https://integrate.api.nvidia.com/v1"
|
| 95 |
+
# )
|
| 96 |
|
| 97 |
agent = create_agent(model, tools)
|
| 98 |
|