arxiv arXiv cs.CL · 1h ago · research

OpenForgeRL enables end-to-end training of harness-based agents in any environment

from English

Researchers present OpenForgeRL, an open-source framework that allows for the end-to-end training of AI agents using complex inference harnesses like Claude Code and OpenClaw. The system decouples training from inference by employing a lightweight proxy to record model calls as data and a Kubernetes orchestrator to manage remote container rollouts.

OpenForgeRL supports diverse environments, including tool-based agents and multimodal GUI browser agents.
OpenForgeClaw achieves 31.7 pass^3 on ClawEval and 33.7 on QwenClawBench using only hundreds to a few thousand tasks.
OpenForgeGUI reaches 37.7 on OSWorld-Verified, 63.0 on Online-Mind2Web, and 72.3 on WebVoyager.
Both variants outperform open baselines of similar size and match or surpass larger models in GUI settings.
Analysis reveals that harness choice significantly impacts learning difficulty and that RL improves reliability metrics like self-verification.

The framework enables researchers to train, study, and improve agents directly within the real harnesses and environments where they are deployed.

Importance 2/3 New harness with differentiators arXiv cs.CL Hugging Face AI agents Training methods

Benchmarks

Benchmark	Model	Score
SWE-bench Pro	OpenForgeClaw	55.9%
OSWorld	OpenForgeGUI	37.7%

Related articles

arxiv arXiv cs.CL · 10d ago

OpenSquilla introduces Harness-Native agentic routing to create a data flywheel

The article proposes Harness-Native agentic routing, a step-level paradigm that selects models based on the full execution harness state rather than optimizing single-turn cost-quality trade-offs. This approach treats every routing decision as a structured data record to form a harness-native data flywheel, where execution traces train better routers and models.

arxiv arXiv cs.AI · 35d ago · 2 views

UltraQuant: 4-bit KV Caching for Context-Heavy Agents

UltraQuant enables 4-bit KV caching for context-heavy agents, reducing P50 time-to-first-token by 3.47x in late rounds and boosting output throughput by 1.63x over FP8 KV baseline. It achieves this using FP8 queries, FP4 KV tensors, UE8M0 group scales, and native scaled-MFMA on AMD CDNA4 GPUs, with optimizations for decode-attention kernels and robust design choices like asymmetric K/V treatment and Walsh-Hadamard rotation.

arxiv arXiv cs.LG · 35d ago · 1 view

UltraQuant: 4-bit KV Caching for Context-Heavy Agents

UltraQuant introduces a 4-bit KV caching method tailored for context-heavy agent workloads. It achieves 3.47x reduction in P50 time-to-first-token in late rounds and 1.63x higher output throughput compared to FP8 KV caching, using FP8 queries, FP4 KV tensors, and native AMD CDNA4 scaled-MFMA support.

arxiv arXiv cs.LG · 35d ago · 1 view

Training LLMs for Long-Lifecycle Agents via Cross-Domain Generalization

A new framework enables large language models to develop 'Connect the Dots' capability, allowing long-lifecycle agents to learn from experiences and iteratively update their environment context. The framework uses reinforcement learning with long rollout sequences and custom tasks to promote cross-domain generalization, showing effective out-of-distribution performance in both domains and transition settings.

arxiv arXiv cs.CL · 35d ago · 1 view

Training LLMs for Long-Lifecycle Agents via Cross-Domain Generalization

A new framework enables large language models to learn 'Connect the Dots' by using reinforcement learning with long rollout sequences. The method includes tailored tasks and environments to foster meta-capability development, showing strong cross-domain generalization and performance in out-of-distribution settings. Implementations are available at https://github.com/agentscope-ai/Trinity-RFT/tree/research/cod/examples/research_cod.