Nitro: Boosting Distributed Reinforcement Learning with Serverless Computing
Summary: Nitro leverages serverless functions to spawn ephemeral actors for instant high-concurrency sampling, avoiding serverful startup and scalability bottlenecks. Using a metric-driven, cost-aware actor-scaling heuristic, Nitro yields up to 6× higher final rewards and 42% lower training cost. (summarized by gpt-5-mini on Feb 09 2026)
Incoming Non-self Citations Over Time
No non-self incoming citations found for this paper in this database.
Authors
- 1. Hanfei Yu
- 2. Jacob Carter
- 3. Hao Wang
- 4. Devesh Tiwari
- 5. Jian Li
- 6. Seung-Jong Park
Incoming Citations (Sorted by Pagerank)
Showing 0 of 0 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 1 of 1 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
| Rank | Cited Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 2,791 | Towards Demystifying Serverless Machine Learning Training | 2021 | SIGMOD | 8.1206618e-05 |
Previous
Page 1 / 1
Next