Training

Training large AI models requires significant compute resources. These resources are concentrated in a few hyperscalers, creating bottlenecks and single points of control. Today we announce the Zoo Compute Network, a decentralized alternative. The Compute Concentration Problem Current AI training is dominated by: Cloud providers: AWS, GCP, Azure control most AI-grade compute Hardware scarcity: H100s have year-long waitlists High costs: Training GPT-4 class models costs $100M+ Geographic concentration: Most clusters are in a few regions This concentration creates risks:...

Training large language models requires more than algorithms. It requires infrastructure: distributed training frameworks, data pipelines, experiment tracking, and evaluation harnesses. Today we open source Training Gym, our complete platform for model development. Why Training Gym? Open AI development faces an infrastructure gap. Publishing model weights is valuable, but it’s not enough. Researchers need: Reproducible training pipelines Scalable distributed training Standardized evaluation Experiment management Data processing tools Training Gym provides all of this in an integrated, open source package....

Reinforcement learning from human feedback (RLHF) has become central to aligning language models with human preferences. But current methods like PPO are sample-inefficient and unstable. Today we introduce Group Relative Policy Optimization (GRPO), a new approach that addresses these limitations. The RLHF Challenge Standard RLHF follows three steps: Train a reward model on human preference data Use the reward model to provide training signal Optimize the policy with reinforcement learning (typically PPO) Step 3 is problematic....

Training large language models requires vast amounts of data. That data often contains sensitive information. Federated learning offers a path to train on distributed, private data without centralizing it. The Centralization Problem Traditional ML training follows a simple pattern: collect data, aggregate it centrally, train models. This creates problems: Privacy risk: Sensitive data leaves user control Legal barriers: Regulations prevent data movement across jurisdictions Trust requirements: Data holders must trust the training party Single points of failure: Central aggregation creates vulnerabilities Federated Learning Basics Federated learning inverts the pattern....

Language models are trained on text. That text represents the accumulated knowledge, reasoning, and creativity of countless individuals. Yet the curation process that selects training data receives surprisingly little attention. The Data Problem Most large language models are trained on web scrapes filtered by simple heuristics. This approach has several issues: Quality variance: Web content ranges from expert research to spam Hidden biases: Filtering decisions embed value judgments Provenance opacity: It’s unclear what’s included or excluded Legal ambiguity: Copyright and consent questions remain unresolved Our Approach: Transparent Curation At Zen, we’re taking a different path....