Jet-Long: Efficient Long-Context Extension with Dynamic Bifocal RoPE

Researchers propose Jet-Long, a tuning-free zero-shot method for extending LLM context windows that dynamically adjusts a rescaling factor based on sequence length, preserving short-context fidelity while enabling long-context extrapolation. On Qwen3 models up to 128K context, Jet-Long outperforms baselines on RULER, HELMET-RAG, and PG-19 benchmarks, and achieves up to 1.39x FA2 throughput on H100 with minimal overhead.

arXiv:2607.07740v1 Announce Type: new Abstract: Modern LLMs are increasingly deployed in long-context applications such as retrieval-augmented generation, repository-level coding, and agentic workflows whose accumulated reasoning and tool traces routinely push the input an order of magnitude past the pretraining window, making zero-shot context extension the dominant deployment path for open-weight checkpoints. Most existing zero-shot methods fix a single rescaling factor up front, so an aggressive factor sacrifices short-context fidelity while a conservative one breaks down at long contexts. We propose Jet-Long, a tuning-free zero-shot method that pairs a local RoPE-faithful window with a long-range window whose rescaling factor adapts dynamically to the current sequence length, recovering the base model exactly at short inputs while extrapolating cleanly at long ones. An inclusion-exclusion attention merge and an on-the-fly RoPE correction rotation make the bifocal construction essentially free at inference; fused into a single CuTe kernel, long-context prefill reaches up to $1.39\times$ FA2 throughput on H100 approaching the Hopper-only FA4 , and single-batch generation incurs $\le 4\%$ overhead at every length. On Qwen3-1.7B/4B/8B up to 128K context, Jet-Long leads RULER by $+4.79$/$+2.18$/$+2.03$~pp over the strongest baseline at 1.7B/4B/8B, achieves the best overall accuracy on HELMET-RAG a benchmark identified by HELMET as the most efficient predictor of downstream long-context performance and attains the lowest PG-19 perplexity. Jet-Long also generalizes to hybrid attention architectures such as Jet-Nemotron for further long-context improvement without retraining, and remains hyperparameter-resilient for ease of deployment.