TechPrufer
Sign inGet started

Blogs›Together AI›Looped Language Models

Looped Language Models

Looped Language Models

3
posts
2025–2026

Together AI is exploring and enabling advanced model architectures like Parcae, which introduce stable looped language models. This approach allows for achieving higher quality with fewer parameters by increasing recurrence rather than solely scaling data, opening efficient frontiers for training memory-constrained on-device models. The research focuses on stabilizing training for these looped architectures, establishing new scaling laws, and demonstrating their effectiveness in matching or exceeding performance of larger models. This partnership with Meta's PyTorch team will integrate sophisticated high-performance Reinforcement Learning (RL) into the Together AI platform, complementing existing services like supervised fine-tuning, sandboxed environments, evaluation tools, model optimizations, and high-performance inference. The goal is to provide a complete platform for building, training, and deploying next-generation agentic systems with advanced RL capabilities.

2026

Parcae: Doing more with fewer parameters using stable looped models

4/15/2026

This post introduces Parcae, a novel stable architecture for looped language models. It details the challenges in training prior looped models (instability, state explosion, loss spikes) and presents a theoretical framework based on linear dynamical systems to understand and address these issues. Parcae achieves stability by parameterizing input injection and constraining the spectral radius of the update matrix. The post empirically demonstrates Parcae's ability to match the quality of larger Transformers with fewer parameters, establishes new scaling laws for looping, and provides training code and models for community use.

CoderForge-Preview: SOTA open dataset for training efficient coding agents

2/25/2026

This post introduces CoderForge-Preview, the largest open dataset of coding agent trajectories (258k test-verified trajectories). It details the data generation process using Qwen3-Coder-480B, including task sources (R2E-Gym, SWE-Smith, SWE-Rebench), setup with OpenHands scaffold, and filtering for test-verified trajectories. The post compares CoderForge-Preview to other datasets, highlighting its scale and long-context coverage. It also analyzes trajectory success rates, data generation costs, trajectory characteristics (length, steps), and license filtering. Furthermore, it describes the training setup for fine-tuning Qwen3-32B using sequence parallelism (Ulysses), multi-packing, and a token-level loss formulation, achieving 59.4% pass@1 on SWE-Bench Verified.

2025

Together AI and Meta partner to bring PyTorch Reinforcement Learning to the AI Native Cloud

12/3/2025

This post announces a partnership between Together AI and Meta's PyTorch team to build an open-source reinforcement learning framework for training agentic AI systems. It highlights the growing importance of RL for modern agentic models that need to operate over longer time horizons and interact with dynamic environments. The integration will bring sophisticated, high-performance RL capabilities to the Together AI platform, complementing existing services and providing a complete solution for building, training, and deploying agentic systems.