Modal Compute Agents

import asyncio
from praisonaiagents import Agent
from praisonai.integrations.managed_agents import ManagedAgent, ManagedConfig

managed = ManagedAgent(
    provider="local",
    config=ManagedConfig(model="gpt-4o-mini"),
    compute="modal"
)

# Deploy function with custom image
info = asyncio.run(managed.provision_compute(
    image="python:3.12",
    packages=["numpy", "scipy", "numba"]  # High-performance packages
))

print(f"Function deployed: {info.instance_id}")
print(f"Status: {info.status}")

# Execute computationally intensive tasks
result = asyncio.run(managed.execute_in_compute("""
python3 -c "
import numpy as np
import time

# Large matrix multiplication
start = time.time()
a = np.random.rand(1000, 1000)
b = np.random.rand(1000, 1000)
c = np.dot(a, b)
end = time.time()

print(f'Matrix multiplication (1000x1000): {end - start:.3f} seconds')
print(f'Result shape: {c.shape}')
print(f'Sample result: {c[0, 0]:.6f}')
"
"""))

print(result["stdout"])

# Execute parallel computations
result = asyncio.run(managed.execute_in_compute("""
python3 -c "
import multiprocessing as mp
import time

def worker(n):
    return sum(i * i for i in range(n))

start = time.time()
with mp.Pool() as pool:
    results = pool.map(worker, [100000] * 8)
end = time.time()

print(f'Parallel computation: {end - start:.3f} seconds')
print(f'Results: {results[:3]}...')
"
"""))

print(result["stdout"])

import asyncio
from praisonaiagents import Agent
from praisonai.integrations.managed_agents import ManagedAgent, ManagedConfig

managed = ManagedAgent(
    provider="local",
    config=ManagedConfig(
        model="gpt-4o-mini",
        system="You are a computational scientist. Use Modal's serverless platform for intensive computations."
    ),
    compute="modal"
)
agent = Agent(name="scientist", backend=managed)

# Provision high-performance environment
info = asyncio.run(managed.provision_compute(
    packages=["numpy", "scipy", "matplotlib", "pandas", "scikit-learn"]
))

# LLM directs high-performance computing
result = agent.start("""
Perform a Monte Carlo simulation to estimate π using 10 million random points.
Show the convergence process and final estimate.
""", stream=True)

import asyncio
from praisonaiagents import Agent
from praisonai.integrations.managed_agents import ManagedAgent, ManagedConfig

managed = ManagedAgent(
    provider="local",
    config=ManagedConfig(
        model="gpt-4o-mini",
        system="You are an ML engineer using Modal for scalable model training."
    ),
    compute="modal"
)
agent = Agent(name="ml-engineer", backend=managed)

# ML environment with GPU support (if available)
info = asyncio.run(managed.provision_compute(
    packages=["torch", "scikit-learn", "numpy", "pandas"]
))

# Train models at scale
result = agent.start("""
Create and train a simple neural network on synthetic data:
1. Generate 10,000 samples with 10 features
2. Create a 3-layer neural network
3. Train for 100 epochs
4. Report final accuracy and training time
""")

import asyncio
from praisonaiagents import Agent
from praisonai.integrations.managed_agents import ManagedAgent, ManagedConfig

managed = ManagedAgent(
    provider="local",
    config=ManagedConfig(
        model="gpt-4o-mini",
        system="You are a data engineer processing large datasets efficiently."
    ),
    compute="modal"
)
agent = Agent(name="data-engineer", backend=managed)

# Big data environment
info = asyncio.run(managed.provision_compute(
    packages=["pandas", "polars", "pyarrow", "dask"]
))

# Process large datasets
result = agent.start("""
Simulate processing a large dataset:
1. Create a 1M row DataFrame with random data
2. Perform groupby operations and aggregations
3. Calculate percentiles and statistics
4. Measure processing time and memory usage
""")

import asyncio
from praisonaiagents import Agent
from praisonai.integrations.managed_agents import ManagedAgent, ManagedConfig

managed = ManagedAgent(
    provider="local",
    config=ManagedConfig(model="gpt-4o-mini"),
    compute="modal"
)
agent = Agent(name="physicist", backend=managed)

# Scientific computing environment
info = asyncio.run(managed.provision_compute(
    packages=["numpy", "scipy", "sympy", "matplotlib"]
))

# Complex calculations
result = agent.start("""
Solve a differential equation numerically:
1. dy/dx = -2*y + x with initial condition y(0) = 1
2. Solve over x = [0, 5]
3. Plot the solution
4. Compare with analytical solution
""")

import asyncio
from praisonaiagents import Agent
from praisonai.integrations.managed_agents import ManagedAgent, ManagedConfig

managed = ManagedAgent(
    provider="local",
    config=ManagedConfig(model="gpt-4o-mini"),
    compute="modal"
)

# GPU-accelerated compute (if available in your Modal plan)
info = asyncio.run(managed.provision_compute(
    packages=["torch", "transformers", "accelerate"],
    gpu="T4",  # or "A10G", "A100" based on availability
    memory=8192  # 8GB for GPU workloads
))

# Use custom Docker images for specialized environments
info = asyncio.run(managed.provision_compute(
    image="tensorflow/tensorflow:latest-gpu",
    packages=["transformers", "datasets"]
))

# Access Modal's shared volumes (if configured)
result = asyncio.run(managed.execute_in_compute("""
# Upload/download files to/from Modal volumes
echo "Data processing complete" > /data/results.txt
cat /data/results.txt
"""))

# Set environment variables for functions
result = asyncio.run(managed.execute_in_compute("""
export COMPUTE_TYPE=modal
export WORKER_ID=$RANDOM
python3 -c "import os; print(f'Running on {os.environ.get(\"COMPUTE_TYPE\")}')"
"""))