The authors built QwenGyre for extreme-long-horizon agent training, where individual rollouts can approach one million tokens.
Elastic GPU scheduling makes reinforcement learning faster for marathon agent tasks
QwenGyre combines live resource reallocation with trajectory deduplication to train agents on rollouts reaching 700,000 tokens.
Chinese Tech
Weiqi Wang · Yuxin Zhou · Mouxiang Chen · Siyuan Zhang · Yi Zhang · Yuyan Luo · +6 more
Alibaba Token Hub, Alibaba Group · University of Science and Technology of China · Tsinghua University
Research Digest··2 min read
Wang et al.
Why this paper
From Alibaba Token Hub, Alibaba Group and 2 others
In one line
QwenGyre enables online reinforcement learning for xlong-horizon LLM agents by elastic GPU reallocation and trajectory deduplication.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ✓Compute or model size stated (params 2.4T)
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§