Spaces:
Sleeping
Sleeping
File size: 8,970 Bytes
c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe c330796 062ecfe | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 | """
OpenCLAW Agent + SEED β HuggingFace Spaces Dashboard
======================================================
Gradio interface with background agent loop and autonomous model growth.
"""
import os
import sys
import json
import threading
import time
import logging
import gradio as gr
from datetime import datetime, timezone
from pathlib import Path
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
from core.config import Config
from core.agent import OpenCLAWAgent, AgentState
from core.strategy import StrategyReflector
from research.arxiv_fetcher import ArxivFetcher
logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s] %(message)s")
logger = logging.getLogger("openclaw")
STATE_DIR = Path(os.getenv("STATE_DIR", "state"))
SEED_STATE_DIR = Path(os.getenv("SEED_STATE_DIR", "seed_state"))
SEED_DATA_DIR = Path(os.getenv("SEED_DATA_DIR", "seed_data"))
agent_running = False
cycle_log = []
seed_log = []
# ==========================================================================
# BACKGROUND AGENT + SEED GROWTH
# ==========================================================================
def run_background_agent():
"""Background thread: agent + SEED growth combined."""
global agent_running, cycle_log, seed_log
agent_running = True
interval = int(os.getenv("DAEMON_INTERVAL", "3600"))
agent = OpenCLAWAgent(state_dir=str(STATE_DIR))
# Initialize SEED
seed_engine = None
try:
from seed.growth_engine import GrowthEngine
seed_engine = GrowthEngine(
hf_token=os.environ.get("HF_TOKEN", ""),
state_dir=str(SEED_STATE_DIR),
data_dir=str(SEED_DATA_DIR),
)
logger.info("π± SEED Growth Engine initialized")
except Exception as e:
logger.warning(f"SEED init failed (will retry): {e}")
cycle_num = 0
while True:
cycle_num += 1
now = datetime.now(timezone.utc).isoformat()
# === AGENT CYCLE ===
try:
results = agent.run_cycle()
entry = f"[{now}] Agent cycle #{cycle_num}: " + ", ".join(
f"{a['task']}={a['status']}" for a in results.get("actions", [])
)
cycle_log.append(entry)
logger.info(entry)
except Exception as e:
cycle_log.append(f"[{now}] Agent error: {e}")
logger.error(f"Agent cycle error: {e}")
# === SEED GROWTH (every 6th cycle = ~6 hours) ===
if seed_engine and cycle_num % 6 == 0:
try:
logger.info("π± Running SEED growth cycle...")
seed_results = seed_engine.run_cycle()
seed_entry = (
f"[{now}] SEED cycle #{seed_results.get('cycle', '?')}: "
f"stage={seed_engine.cycle_log.get('current_stage', '?')}, "
f"data={seed_engine.cycle_log.get('total_data_harvested', 0)}"
)
seed_log.append(seed_entry)
logger.info(seed_entry)
except Exception as e:
seed_log.append(f"[{now}] SEED error: {e}")
logger.error(f"SEED cycle error: {e}")
elif seed_engine is None:
# Retry SEED init
try:
from seed.growth_engine import GrowthEngine
seed_engine = GrowthEngine(
hf_token=os.environ.get("HF_TOKEN", ""),
state_dir=str(SEED_STATE_DIR),
data_dir=str(SEED_DATA_DIR),
)
except Exception:
pass
# Keep logs bounded
cycle_log[:] = cycle_log[-200:]
seed_log[:] = seed_log[-100:]
time.sleep(interval)
# Start background
bg_thread = threading.Thread(target=run_background_agent, daemon=True)
bg_thread.start()
logger.info("π Background agent + SEED started")
# ==========================================================================
# DASHBOARD FUNCTIONS
# ==========================================================================
def get_status():
state_file = STATE_DIR / "agent_state.json"
if state_file.exists():
try:
state = json.loads(state_file.read_text())
return json.dumps(state, indent=2)
except Exception:
pass
return json.dumps({
"status": "running" if agent_running else "starting",
"message": "Agent initializing...",
}, indent=2)
def get_activity():
return "\n".join(cycle_log[-50:]) if cycle_log else "No activity yet β first cycle in progress..."
def get_papers():
fetcher = ArxivFetcher()
papers = fetcher.known_papers
lines = []
for p in papers:
lines.append(f"π {p['title']}")
lines.append(f" ID: {p['arxiv_id']} | Year: {p.get('year', '?')}")
lines.append(f" {p.get('abstract', '')[:150]}...")
lines.append("")
return "\n".join(lines) if lines else "No papers loaded yet."
def get_strategy():
reflector = StrategyReflector(state_dir=str(STATE_DIR))
report = reflector.analyze()
return json.dumps(report, indent=2)
def get_seed_status():
"""Get SEED growth status."""
try:
from seed.growth_engine import GrowthEngine
engine = GrowthEngine(
hf_token=os.environ.get("HF_TOKEN", ""),
state_dir=str(SEED_STATE_DIR),
data_dir=str(SEED_DATA_DIR),
)
status = engine.get_status()
return json.dumps(status, indent=2, default=str)
except Exception as e:
return json.dumps({"status": "initializing", "note": str(e)}, indent=2)
def get_seed_log():
return "\n".join(seed_log[-50:]) if seed_log else "SEED not yet active β first growth cycle pending..."
def trigger_harvest():
"""Manually trigger data harvest."""
try:
from seed.data.harvester import DataHarvester
h = DataHarvester(str(SEED_DATA_DIR))
stats = h.harvest_all()
return f"β
Harvested {stats['total']} entries:\n" + json.dumps(stats, indent=2)
except Exception as e:
return f"β Harvest failed: {e}"
def trigger_cycle():
try:
agent = OpenCLAWAgent(state_dir=str(STATE_DIR))
results = agent.run_cycle()
return json.dumps(results, indent=2, default=str)
except Exception as e:
return f"Error: {e}"
# ==========================================================================
# GRADIO INTERFACE
# ==========================================================================
with gr.Blocks(title="π± OpenCLAW SEED β Self-Evolving Agent", theme=gr.themes.Soft()) as demo:
gr.Markdown("""
# π± OpenCLAW SEED β Self-Evolving Autonomous Agent
*A seed that grows into a tree. Autonomous 24/7 research agent with self-training AI.*
**By Francisco Angulo de Lafuente** | [GitHub](https://github.com/Agnuxo1) | [Scholar](https://scholar.google.com/citations?user=6nOpJ9IAAAAJ)
""")
with gr.Tab("π Agent Status"):
status_box = gr.Code(label="Agent State", language="json")
gr.Button("Refresh").click(get_status, outputs=status_box)
demo.load(get_status, outputs=status_box)
with gr.Tab("π Activity Log"):
log_box = gr.Textbox(label="Recent Activity", lines=20, max_lines=30)
gr.Button("Refresh").click(get_activity, outputs=log_box)
demo.load(get_activity, outputs=log_box)
with gr.Tab("π± SEED Growth"):
gr.Markdown("""
### Self-Evolving Model Growth
SEED harvests knowledge, trains itself, and grows autonomously.
**Growth stages:** GERMINATION (0.5B) β SEEDLING (1B) β SAPLING (3B) β YOUNG_TREE (7B) β MATURE_TREE (13B+)
""")
seed_status_box = gr.Code(label="SEED Status", language="json")
seed_log_box = gr.Textbox(label="Growth Log", lines=10)
with gr.Row():
gr.Button("π Refresh Status").click(get_seed_status, outputs=seed_status_box)
gr.Button("πΎ Harvest Data Now").click(trigger_harvest, outputs=seed_log_box)
demo.load(get_seed_status, outputs=seed_status_box)
with gr.Tab("π Research Papers"):
papers_box = gr.Textbox(label="Known Papers", lines=20, max_lines=30)
gr.Button("Refresh").click(get_papers, outputs=papers_box)
demo.load(get_papers, outputs=papers_box)
with gr.Tab("π§ Strategy"):
strategy_box = gr.Code(label="Strategy Analysis", language="json")
gr.Button("Analyze").click(get_strategy, outputs=strategy_box)
with gr.Tab("β‘ Manual Trigger"):
trigger_box = gr.Code(label="Cycle Results", language="json")
gr.Button("π€ Run Agent Cycle").click(trigger_cycle, outputs=trigger_box)
if __name__ == "__main__":
port = int(os.environ.get("PORT", 7860))
demo.launch(server_name="0.0.0.0", server_port=port)
|