Spaces:
Running
on
CPU Upgrade
Running
on
CPU Upgrade
File size: 9,032 Bytes
d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 d1ed69b 3119795 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 |
import os
import sys
import pathlib
import shutil
import threading
import multiprocessing
import io
import yaml
import gradio as gr
from loguru import logger
from yourbench.pipeline import run_pipeline
UPLOAD_DIRECTORY = pathlib.Path("/app/uploaded_files")
UPLOAD_DIRECTORY.mkdir(parents=True, exist_ok=True)
CONFIG_PATH = pathlib.Path("/app/yourbench_config.yml")
logger.remove()
logger.add(sys.stderr, level="INFO")
import subprocess
import io
import os
import time
class SubprocessManager:
def __init__(self, command):
self.command = command
self.process = None
self.output_stream = io.StringIO()
def start_process(self):
"""Start the subprocess."""
if self.is_running():
logger.info("Process is already running")
return
self.process = subprocess.Popen(
self.command,
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT, # Combine stderr with stdout
text=True,
bufsize=1, # Line-buffered
start_new_session=True # Start the process in a new session
)
os.set_blocking(self.process.stdout.fileno(), False)
logger.info("Started the process")
def read_and_get_output(self):
"""Read available subprocess output and return the captured output."""
if self.process and self.process.stdout:
try:
while True:
line = self.process.stdout.readline()
if line:
self.output_stream.write(line) # Capture in StringIO
else:
break
except BlockingIOError:
pass
return self.output_stream.getvalue()
def stop_process(self):
"""Terminate the subprocess."""
if not self.is_running():
logger.info("Started the process")
return
logger.info("Sending SIGTERM to the Process")
self.process.terminate()
exit_code = self.process.wait() # Wait for process to terminate
logger.info(f"Process stopped exit code {exit_code}")
#return exit_code
def kill_process(self):
"""Forcefully kill the subprocess."""
if not self.is_running():
logger.info("Process is not running")
return
logger.info("Sending SIGKILL to the Process")
self.process.kill()
exit_code = self.process.wait() # Wait for process to be killed
logger.info(f"Process killed exit code {exit_code}")
#return exit_code
def is_running(self):
"""Check if the subprocess is still running."""
return self.process and self.process.poll() is None
command = ["uv", "run", "yourbench", f"--config={CONFIG_PATH}"]
manager = SubprocessManager(command)
def generate_config(hf_token, hf_org, model_name, provider, base_url, api_key, max_concurrent_requests):
config = {
"hf_configuration": {
"token": hf_token,
"private": True,
"hf_organization": hf_org
},
"model_list": [{
"model_name": model_name,
"provider": provider,
"base_url": base_url,
"api_key": api_key,
"max_concurrent_requests": max_concurrent_requests
}],
"model_roles": {role: [model_name] for role in [
"ingestion", "summarization", "single_shot_question_generation",
"multi_hop_question_generation", "answer_generation", "judge_answers"
]},
"inference_config": {"max_concurrent_requests": 16},
"pipeline": {
"ingestion": {
"source_documents_dir": "/app/uploaded_files",
"output_dir": "/app/ingested",
"run": True
},
"upload_ingest_to_hub": {
"source_documents_dir": "/app/ingested",
"hub_dataset_name": "test_ingested_documents",
"local_dataset_path": "/app/ingested_dataset",
"run": True
},
"summarization": {
"source_dataset_name": "test_ingested_documents",
"output_dataset_name": "test_summaries",
"local_dataset_path": "/results/test_summaries",
"concat_existing_dataset": False,
"run": True
},
"chunking": {
"source_dataset_name": "test_summaries",
"output_dataset_name": "test_chunked_documents",
"local_dataset_path": "/results/test_chunked_documents",
"concat_existing_dataset": False,
"chunking_configuration": {
"l_min_tokens": 64,
"l_max_tokens": 128,
"tau_threshold": 0.3,
"h_min": 2,
"h_max": 4
},
"run": True
},
"single_shot_question_generation": {
"source_dataset_name": "test_chunked_documents",
"output_dataset_name": "test_single_shot_questions",
"local_dataset_path": "/results/test_single_shot_questions",
"diversification_seed": "24 year old adult",
"concat_existing_dataset": False,
"run": True
},
"multi_hop_question_generation": {
"source_dataset_name": "test_chunked_documents",
"output_dataset_name": "test_multi_hop_questions",
"local_dataset_path": "/results/test_multi_hop_questions",
"concat_existing_dataset": False,
"run": True
},
"answer_generation": {
"run": True,
"question_dataset_name": "test_single_shot_questions",
"output_dataset_name": "test_answered_questions",
"local_dataset_path": "/results/test_answered_questions",
"concat_existing_dataset": False,
"strategies": [{
"name": "zeroshot",
"prompt": "ZEROSHOT_QA_USER_PROMPT",
"model_name": model_name
}, {
"name": "gold",
"prompt": "GOLD_QA_USER_PROMPT",
"model_name": model_name
}]
},
"judge_answers": {
"run": True,
"source_judge_dataset_name": "test_answered_questions",
"output_judged_dataset_name": "test_judged_comparisons",
"local_dataset_path": "/results/test_judged_comparisons",
"concat_existing_dataset": False,
"comparing_strategies": [["zeroshot", "gold"]],
"chunk_column_index": 0,
"random_seed": 42
}
}
}
return yaml.dump(config, default_flow_style=False)
def save_config(yaml_text):
with open(CONFIG_PATH, "w") as file:
file.write(yaml_text)
return "✅ Config saved!"
def save_files(files: list[str]):
saved_paths = [shutil.move(str(pathlib.Path(file)), str(UPLOAD_DIRECTORY / pathlib.Path(file).name)) for file in files]
return f"Files saved to: {', '.join(saved_paths)}"
app = gr.Blocks()
with app:
gr.Markdown("## YourBench Configuration")
with gr.Tab("Configuration"):
hf_token = gr.Textbox(label="HF Token")
hf_org = gr.Textbox(label="HF Organization")
model_name = gr.Textbox(label="Model Name")
provider = gr.Dropdown(["openrouter", "openai", "huggingface"], value="huggingface", label="Provider")
base_url = gr.Textbox(label="Base URL")
api_key = gr.Textbox(label="API Key")
max_concurrent_requests = gr.Dropdown([8, 16, 32], value=16, label="Max Concurrent Requests")
config_output = gr.Code(label="Generated Config", language="yaml")
preview_button = gr.Button("Generate Config")
save_button = gr.Button("Save Config")
preview_button.click(generate_config, inputs=[hf_token, hf_org, model_name, provider, base_url, api_key, max_concurrent_requests], outputs=config_output)
save_button.click(save_config, inputs=[config_output], outputs=[gr.Textbox(label="Save Status")])
with gr.Tab("Files"):
file_input = gr.File(label="Upload text files", file_count="multiple", file_types=[".txt", ".md", ".html"])
output = gr.Textbox(label="Log")
file_input.upload(save_files, file_input, output)
with gr.Tab("Run Generation"):
log_output = gr.Code(label="Log Output", language=None, lines=20, interactive=False)
start_button = gr.Button("Start Task")
start_button.click(manager.start_process)
timer = gr.Timer(0.1, active=True)
timer.tick(manager.read_and_get_output, outputs=log_output)
start_button = gr.Button("Kill Task")
start_button.click(manager.kill_process)
app.launch() |