Sprint en vivo · Hora 00 → 48

48h Climate Data Sprint Inspira. Actúa. Transforma.

Datos reales de clima y energía de toda Latinoamérica. Build the model, surface the pattern, and tell the story that moves policy.

10,000
Datos → records
T4 · $20
Modelos → compute + LLM
47:58:12
Impacto → deadline
Kickoff
Datos liberados
Sprint
Activo ahora
Pitch & Judge
Hora 48
Climate Impact
Insights → acción

Bienvenidos, Team SolarPulse

Rank #8 / 50
JupyterHub URL
p3x9k.proxy.runpod.net
OpenRouter Key
sk-or-v1-a8f3…d92e
Dataset · HuggingFace
cleantechhub/datathon-2025

Se les entrega un dataset de 10,000 registros de consumo energético de municipios de Latinoamérica — variables climáticas, datos demográficos y de política. Predigan el potencial de reducción de CO₂ (kg/año) alcanzable mediante transiciones a energía renovable.

Métrica: R² — más alto es mejor.  Formato: CSV con columnas id, prediction.  Límite: 10 envíos por día.

Your GPU Notebook

Online
JupyterLab 4.1 · Python 3.11 · CUDA 12.1
# Quick start — run this in your first cell
import os, datasets, pandas as pd

ds = datasets.load_dataset("cleantechhub/datathon-2025")
df_train = ds["train"].to_pandas()
df_test  = ds["test"].to_pandas()

import torch
print(f"GPU: {torch.cuda.get_device_name(0)}")
T4
GPU model
16 GB
VRAM
100 GB
SSD storage
Open JupyterLab in new tab

LLM API · OpenRouter

Cap $20.00
Your API key · hard cap $20.00
sk-or-v1-a8f3c7b2e1d4f9a0c5b8e3d92e
example.py
from openai import OpenAI
client = OpenAI(api_key="sk-or-v1-…", base_url="https://openrouter.ai/api/v1")
resp = client.chat.completions.create(
    model="anthropic/claude-3-5-sonnet",
    messages=[{"role":"user", "content": f"CO₂ potential: {row}"}])
Probar el API en vivo
Sends one sample row to claude-3-5-sonnet and draws down your budget.
ModelIn / 1MOut / 1MBest for
claude-3-5-sonnet$3.00$15.00Analysis, reasoning
gpt-4o$2.50$10.00Code, structured output
gemini-pro-1.5$1.25$5.00Long context
llama-3.1-70b$0.88$0.88Fast, budget
mistral-large$2.00$6.00Multilingual

Submit Your Solution

8 / 10 today

Upload predictions as CSV with columns id, prediction. Scoring runs automatically on a held-out set via Modal — results land on the leaderboard in ~3 min.

Drop your CSV here or click to browse

Max 50 MB · columns [id, prediction]


#TimeScore (R²)Status

Live Leaderboard

Updates 3 min
You · #8
#TeamScore (R²)Subs