Spaces:
Sleeping
Sleeping
fix code
Browse files
app.py
CHANGED
|
@@ -8,9 +8,12 @@ import gradio as gr
|
|
| 8 |
import pandas as pd
|
| 9 |
import requests
|
| 10 |
|
| 11 |
-
from smolagents import CodeAgent, tool
|
| 12 |
-
|
| 13 |
-
|
|
|
|
|
|
|
|
|
|
| 14 |
|
| 15 |
|
| 16 |
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
|
|
@@ -18,29 +21,11 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
|
|
| 18 |
|
| 19 |
@tool
|
| 20 |
def get_task_file_url(task_id: str) -> str:
|
| 21 |
-
"""
|
| 22 |
-
Returns the download URL for the file attached to a GAIA task.
|
| 23 |
-
|
| 24 |
-
Args:
|
| 25 |
-
task_id: The GAIA task id.
|
| 26 |
-
|
| 27 |
-
Returns:
|
| 28 |
-
Direct URL to the attached file for this task.
|
| 29 |
-
"""
|
| 30 |
return f"{DEFAULT_API_URL}/files/{task_id}"
|
| 31 |
|
| 32 |
|
| 33 |
@tool
|
| 34 |
def download_file(url: str) -> str:
|
| 35 |
-
"""
|
| 36 |
-
Downloads a file from a URL to a temporary local path and returns that path.
|
| 37 |
-
|
| 38 |
-
Args:
|
| 39 |
-
url: The file URL.
|
| 40 |
-
|
| 41 |
-
Returns:
|
| 42 |
-
Local file path as a string.
|
| 43 |
-
"""
|
| 44 |
suffix = ""
|
| 45 |
match = re.search(r"\.([a-zA-Z0-9]{1,6})(?:\?|$)", url)
|
| 46 |
if match:
|
|
@@ -60,16 +45,6 @@ def download_file(url: str) -> str:
|
|
| 60 |
|
| 61 |
@tool
|
| 62 |
def inspect_local_text_file(path: str, max_chars: int = 12000) -> str:
|
| 63 |
-
"""
|
| 64 |
-
Reads a local text-like file and returns a preview.
|
| 65 |
-
|
| 66 |
-
Args:
|
| 67 |
-
path: Local file path.
|
| 68 |
-
max_chars: Max number of characters to return.
|
| 69 |
-
|
| 70 |
-
Returns:
|
| 71 |
-
Text preview.
|
| 72 |
-
"""
|
| 73 |
p = Path(path)
|
| 74 |
try:
|
| 75 |
return p.read_text(encoding="utf-8", errors="ignore")[:max_chars]
|
|
@@ -79,17 +54,6 @@ def inspect_local_text_file(path: str, max_chars: int = 12000) -> str:
|
|
| 79 |
|
| 80 |
@tool
|
| 81 |
def analyze_spreadsheet(path: str, instruction: str) -> str:
|
| 82 |
-
"""
|
| 83 |
-
Loads an Excel/CSV spreadsheet with pandas and executes a constrained analysis
|
| 84 |
-
instruction by giving the LLM a schema + preview.
|
| 85 |
-
|
| 86 |
-
Args:
|
| 87 |
-
path: Local spreadsheet path.
|
| 88 |
-
instruction: What to compute.
|
| 89 |
-
|
| 90 |
-
Returns:
|
| 91 |
-
JSON string with columns, preview, dtypes, shape.
|
| 92 |
-
"""
|
| 93 |
p = Path(path)
|
| 94 |
if p.suffix.lower() in [".csv", ".tsv"]:
|
| 95 |
sep = "\t" if p.suffix.lower() == ".tsv" else ","
|
|
@@ -113,15 +77,6 @@ def analyze_spreadsheet(path: str, instruction: str) -> str:
|
|
| 113 |
|
| 114 |
@tool
|
| 115 |
def python_compute(code: str) -> str:
|
| 116 |
-
"""
|
| 117 |
-
Executes trusted short Python snippets for local analysis.
|
| 118 |
-
|
| 119 |
-
Args:
|
| 120 |
-
code: Python code. Must assign the final value to a variable named `result`.
|
| 121 |
-
|
| 122 |
-
Returns:
|
| 123 |
-
Stringified result.
|
| 124 |
-
"""
|
| 125 |
local_vars = {}
|
| 126 |
global_vars = {
|
| 127 |
"pd": pd,
|
|
@@ -147,7 +102,6 @@ def clean_final_answer(text: str) -> str:
|
|
| 147 |
answer = re.sub(r"^(FINAL ANSWER\s*:\s*)", "", answer, flags=re.IGNORECASE)
|
| 148 |
answer = re.sub(r"^(Answer\s*:\s*)", "", answer, flags=re.IGNORECASE)
|
| 149 |
answer = re.sub(r"[ \t]+", " ", answer).strip()
|
| 150 |
-
|
| 151 |
return answer
|
| 152 |
|
| 153 |
|
|
@@ -162,8 +116,7 @@ Rules:
|
|
| 162 |
- For exact-match scoring, your final output must be ONLY the answer and nothing else.
|
| 163 |
- Do not include explanations in your final answer.
|
| 164 |
- Do not include the words FINAL ANSWER.
|
| 165 |
-
- When a question requests a format
|
| 166 |
-
obey that exact format.
|
| 167 |
- If the answer is numeric and the prompt requests decimals, format correctly.
|
| 168 |
"""
|
| 169 |
|
|
@@ -184,22 +137,37 @@ class GaiaAgent:
|
|
| 184 |
max_tokens=1200,
|
| 185 |
)
|
| 186 |
|
| 187 |
-
|
| 188 |
-
|
| 189 |
-
|
| 190 |
-
|
| 191 |
-
|
| 192 |
-
|
| 193 |
-
|
| 194 |
-
|
| 195 |
-
|
| 196 |
-
|
| 197 |
-
|
| 198 |
-
|
| 199 |
-
|
| 200 |
-
|
| 201 |
-
|
| 202 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 203 |
|
| 204 |
def __call__(self, question: str, task_id: str) -> str:
|
| 205 |
full_prompt = f"""
|
|
@@ -316,6 +284,5 @@ with gr.Blocks() as demo:
|
|
| 316 |
outputs=[status_output, results_table],
|
| 317 |
)
|
| 318 |
|
| 319 |
-
|
| 320 |
if __name__ == "__main__":
|
| 321 |
demo.launch(debug=True)
|
|
|
|
| 8 |
import pandas as pd
|
| 9 |
import requests
|
| 10 |
|
| 11 |
+
from smolagents import CodeAgent, tool, DuckDuckGoSearchTool, VisitWebpageTool
|
| 12 |
+
|
| 13 |
+
try:
|
| 14 |
+
from smolagents import OpenAIServerModel
|
| 15 |
+
except ImportError:
|
| 16 |
+
from smolagents import OpenAIModel as OpenAIServerModel
|
| 17 |
|
| 18 |
|
| 19 |
DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
|
|
|
|
| 21 |
|
| 22 |
@tool
|
| 23 |
def get_task_file_url(task_id: str) -> str:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
return f"{DEFAULT_API_URL}/files/{task_id}"
|
| 25 |
|
| 26 |
|
| 27 |
@tool
|
| 28 |
def download_file(url: str) -> str:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
suffix = ""
|
| 30 |
match = re.search(r"\.([a-zA-Z0-9]{1,6})(?:\?|$)", url)
|
| 31 |
if match:
|
|
|
|
| 45 |
|
| 46 |
@tool
|
| 47 |
def inspect_local_text_file(path: str, max_chars: int = 12000) -> str:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
p = Path(path)
|
| 49 |
try:
|
| 50 |
return p.read_text(encoding="utf-8", errors="ignore")[:max_chars]
|
|
|
|
| 54 |
|
| 55 |
@tool
|
| 56 |
def analyze_spreadsheet(path: str, instruction: str) -> str:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
p = Path(path)
|
| 58 |
if p.suffix.lower() in [".csv", ".tsv"]:
|
| 59 |
sep = "\t" if p.suffix.lower() == ".tsv" else ","
|
|
|
|
| 77 |
|
| 78 |
@tool
|
| 79 |
def python_compute(code: str) -> str:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 80 |
local_vars = {}
|
| 81 |
global_vars = {
|
| 82 |
"pd": pd,
|
|
|
|
| 102 |
answer = re.sub(r"^(FINAL ANSWER\s*:\s*)", "", answer, flags=re.IGNORECASE)
|
| 103 |
answer = re.sub(r"^(Answer\s*:\s*)", "", answer, flags=re.IGNORECASE)
|
| 104 |
answer = re.sub(r"[ \t]+", " ", answer).strip()
|
|
|
|
| 105 |
return answer
|
| 106 |
|
| 107 |
|
|
|
|
| 116 |
- For exact-match scoring, your final output must be ONLY the answer and nothing else.
|
| 117 |
- Do not include explanations in your final answer.
|
| 118 |
- Do not include the words FINAL ANSWER.
|
| 119 |
+
- When a question requests a format, obey that exact format.
|
|
|
|
| 120 |
- If the answer is numeric and the prompt requests decimals, format correctly.
|
| 121 |
"""
|
| 122 |
|
|
|
|
| 137 |
max_tokens=1200,
|
| 138 |
)
|
| 139 |
|
| 140 |
+
try:
|
| 141 |
+
self.agent = CodeAgent(
|
| 142 |
+
model=self.model,
|
| 143 |
+
tools=[
|
| 144 |
+
DuckDuckGoSearchTool(),
|
| 145 |
+
VisitWebpageTool(),
|
| 146 |
+
get_task_file_url,
|
| 147 |
+
download_file,
|
| 148 |
+
inspect_local_text_file,
|
| 149 |
+
analyze_spreadsheet,
|
| 150 |
+
python_compute,
|
| 151 |
+
],
|
| 152 |
+
add_base_tools=False,
|
| 153 |
+
max_steps=10,
|
| 154 |
+
instructions=SYSTEM_PROMPT,
|
| 155 |
+
)
|
| 156 |
+
except TypeError:
|
| 157 |
+
self.agent = CodeAgent(
|
| 158 |
+
model=self.model,
|
| 159 |
+
tools=[
|
| 160 |
+
DuckDuckGoSearchTool(),
|
| 161 |
+
VisitWebpageTool(),
|
| 162 |
+
get_task_file_url,
|
| 163 |
+
download_file,
|
| 164 |
+
inspect_local_text_file,
|
| 165 |
+
analyze_spreadsheet,
|
| 166 |
+
python_compute,
|
| 167 |
+
],
|
| 168 |
+
add_base_tools=False,
|
| 169 |
+
max_steps=10,
|
| 170 |
+
)
|
| 171 |
|
| 172 |
def __call__(self, question: str, task_id: str) -> str:
|
| 173 |
full_prompt = f"""
|
|
|
|
| 284 |
outputs=[status_output, results_table],
|
| 285 |
)
|
| 286 |
|
|
|
|
| 287 |
if __name__ == "__main__":
|
| 288 |
demo.launch(debug=True)
|