ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning | Paper Citation Record

Observation dff35b60-e754-4dac-a87f-5f0fbb4a0890 · outbound

This paper cites Outrageously large neural net- works: The sparsely-gated mixture-of-experts layer.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Outrageously large neural net- works: The sparsely-gated mixture-of-experts layer

Reference 1

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.999101Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 52b4106b-93ce-4111-9933-d6c751b7684a · outbound

This paper cites Gshard: Scaling giant models with conditional computation and auto- matic sharding.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Gshard: Scaling giant models with conditional computation and auto- matic sharding

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:59.024843Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation f179413e-f69c-4a37-9723-4df3e915bebe · outbound

This paper cites Switch transform- ers: Scaling to trillion parameter models with simple and efficient sparsity.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Switch transform- ers: Scaling to trillion parameter models with simple and efficient sparsity

Reference 3

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:59.016199Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation ecd82fc7-d778-4c60-91dd-5f0394840717 · outbound

This paper cites Introducing DBRX: A New State-of-the-Art Open LLM.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Introducing DBRX: A New State-of-the-Art Open LLM

Reference 4

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:59.004084Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 46044026-8da8-4b38-bebe-02303127681c · outbound

This paper cites Mixtral of Experts.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Mixtral of Experts

Reference 5

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.870115Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 85e6d9d5-5143-4400-88ae-828e652f1cf6 · outbound

This paper cites DeepSeek-V3 Technical Report.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning DeepSeek-V3 Technical Report

Reference 6

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.867213Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation af418330-8429-4356-97e2-f18e217f2f0c · outbound

This paper cites Qwen3 Technical Report.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Qwen3 Technical Report

Reference 7

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.876080Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation ef79a0e6-6687-4f80-9749-e6e0136ae0d7 · outbound

This paper cites Kimi K2: Open Agentic Intelligence.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Kimi K2: Open Agentic Intelligence

Reference 8

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.873230Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation b9beb139-041a-48c5-8ca1-e0ea276e2ab6 · outbound

This paper cites MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention

Reference 9

Resolution: verified exact; arxiv_id, observed 2026-05-12T09:28:16.586976Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 4ff822f2-3ccd-4f9e-bc5b-7bb03601c371 · outbound

This paper cites Glam: Efficient scaling of language models with mixture-of- experts.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Glam: Efficient scaling of language models with mixture-of- experts

Reference 10

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:59.020714Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 21a61d10-097b-4f92-9226-519e5e265d26 · outbound

This paper cites Deepspeed- moe: Advancing mixture-of-experts inference and train- ing to power next-generation ai scale.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Deepspeed- moe: Advancing mixture-of-experts inference and train- ing to power next-generation ai scale

Reference 11

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.994171Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 568f501b-5448-4eeb-8040-19bb1e4d07bd · outbound

This paper cites Training language models to follow instructions with human feedback.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Training language models to follow instructions with human feedback

Reference 12

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:59.011896Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 71d13990-1ed4-4027-9c1e-577e84dad624 · outbound

This paper cites Deepseek-r1 incentivizes reasoning in llms through reinforcement learning.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Deepseek-r1 incentivizes reasoning in llms through reinforcement learning

Reference 13

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.989243Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 77bbf96c-60d6-4a09-8f5b-296fff3f7779 · outbound

This paper cites Flexmoe: Scaling large-scale sparse pre-trained model training via dynamic device placement.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Flexmoe: Scaling large-scale sparse pre-trained model training via dynamic device placement

Reference 14

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.963005Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation deb4980a-5017-43e3-991b-5a59f9b5f5e7 · outbound

This paper cites {SmartMoE}: Efficiently training {Sparsely- Activated} models through combining offline and online parallelization.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning {SmartMoE}: Efficiently training {Sparsely- Activated} models through combining offline and online parallelization

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:59.008240Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 7b2f7151-8195-4cec-9a5e-1b88bcb32ff8 · outbound

This paper cites Micromoe: Fine- grained load balancing for mixture-of-experts with token scheduling.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Micromoe: Fine- grained load balancing for mixture-of-experts with token scheduling

Reference 16

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.773337Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation f9ad5cd0-4e1a-4ecf-b654-6c62915a71ab · outbound

This paper cites {PopFetcher}: Towards acceler- ated {Mixture-of-Experts} training via popularity based {Expert-Wise}prefetch.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning {PopFetcher}: Towards acceler- ated {Mixture-of-Experts} training via popularity based {Expert-Wise}prefetch

Reference 17

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.884605Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 5bd5b82c-3c4c-4369-a810-19e5545c25f0 · outbound

This paper cites Laer-moe: Load-adaptive expert re-layout for efficient mixture-of-experts training.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Laer-moe: Load-adaptive expert re-layout for efficient mixture-of-experts training

Reference 18

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.935971Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 385433df-5e23-49b7-8fc0-dbd1ef95d4af · outbound

This paper cites Moe parallel folding: Heterogeneous parallelism mappings for efficient large-scale moe model training with megatron core.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Moe parallel folding: Heterogeneous parallelism mappings for efficient large-scale moe model training with megatron core

Reference 19

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.788840Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation ce6e0d2e-58a2-4e5f-a990-bdaa7bf10f97 · outbound

This paper cites DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models

Reference 20

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.863718Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation e6dc883e-635b-4b41-9755-5044ab1dc1b7 · outbound

This paper cites DAPO: An Open-Source LLM Reinforcement Learning System at Scale.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning DAPO: An Open-Source LLM Reinforcement Learning System at Scale

Reference 21

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.776908Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 1b77baf5-dc1f-4d01-8e6a-f27350964c5e · outbound

This paper cites Competition-level code generation with alphacode.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Competition-level code generation with alphacode

Reference 22

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.794293Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 1be9df3d-fb67-4348-aab2-2fbd64496764 · outbound

This paper cites Generalizing Verifiable Instruction Following.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Generalizing Verifiable Instruction Following

Reference 23

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.784748Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 2c3bc7b3-94fa-49e3-a0a5-3fbd26ae892d · outbound

This paper cites DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models

Reference 24

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.852010Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 5d5d2efa-4075-4991-9e0b-cf73ed260b87 · outbound

This paper cites Expert Parallelism Load Balancer (EPLB).

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Expert Parallelism Load Balancer (EPLB)

Reference 25

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.880483Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 624d296b-b2b6-4098-be53-8a4a9e9998ae · outbound

This paper cites Human-level control through deep reinforcement learning.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Human-level control through deep reinforcement learning

Reference 26

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.931027Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 857db13a-d183-4cc8-8bb7-a5ce7a4eaac1 · outbound

This paper cites Soft actor-critic: Off-policy maximum entropy deep rein- forcement learning with a stochastic actor.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Soft actor-critic: Off-policy maximum entropy deep rein- forcement learning with a stochastic actor

Reference 27

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.940727Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation dc74dbbc-b1de-43a9-bfa9-aa9bd083e354 · outbound

This paper cites Stabilizing MoE Reinforcement Learning by Aligning Training and Inference Routers.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Stabilizing MoE Reinforcement Learning by Aligning Training and Inference Routers

Reference 28

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.805376Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation a6f72534-1f13-44d6-9a98-16638ee235e6 · outbound

This paper cites Stabilizing reinforcement learning with llms: Formulation and practices.arXiv preprint arXiv:2512.01374, 2025a.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Stabilizing reinforcement learning with llms: Formulation and practices.arXiv preprint arXiv:2512.01374, 2025a

Reference 29

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.833281Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation f2f9a668-a7c0-4f77-b354-b157fbe567a3 · outbound

This paper cites Optimization and approximation in determinis- tic sequencing and scheduling: a survey.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Optimization and approximation in determinis- tic sequencing and scheduling: a survey

Reference 30

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.945201Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 8e7af441-3f0e-408b-99fb-28eb637f96a2 · outbound

This paper cites At- tention is all you need.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning At- tention is all you need

Reference 31

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.807726Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 18f9c5e4-fdf6-46e1-adec-e442c670b004 · outbound

This paper cites Zero: Memory optimizations toward training trillion parame- ter models.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Zero: Memory optimizations toward training trillion parame- ter models

Reference 32

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.798827Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 2fd505dd-cf8f-468e-a56b-0ff75cc59c7c · outbound

This paper cites Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism

Reference 33

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.855673Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 06914383-ba02-41d5-938b-202de464d0a3 · outbound

This paper cites Gpipe: Efficient training of giant neural networks using pipeline parallelism.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Gpipe: Efficient training of giant neural networks using pipeline parallelism

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.917601Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 32c4b5cc-8779-46e3-b680-9b2c93e941dd · outbound

This paper cites Pipedream: Generalized pipeline parallelism for dnn training.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Pipedream: Generalized pipeline parallelism for dnn training

Reference 35

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.926894Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation bb3e96d9-2d93-4f65-9896-eca7ea46162c · outbound

This paper cites Gpqa: A graduate-level google-proof q&a benchmark.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Gpqa: A graduate-level google-proof q&a benchmark

Reference 36

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.912608Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 65345a14-10bf-4d85-9117-c210ec1ff531 · outbound

This paper cites Orca: A distributed serving system for {Transformer-Based} generative models.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Orca: A distributed serving system for {Transformer-Based} generative models

Reference 37

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.903645Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 9317bfce-1349-4e73-8f1d-47f89af2593f · outbound

This paper cites Fast Distributed Inference Serving for Large Language Models.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Fast Distributed Inference Serving for Large Language Models

Reference 38

Resolution: verified exact; arxiv_id, observed 2026-05-17T11:23:46.844514Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 9fb964f2-6b02-4965-8b77-0340e6bd5fd0 · outbound

This paper cites Proximal Policy Optimization Algorithms.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Proximal Policy Optimization Algorithms

Reference 39

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.829234Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation e3303144-1b90-4337-984c-f1e09be065ac · outbound

This paper cites Taming the long-tail: Efficient reasoning rl training with adaptive drafter.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Taming the long-tail: Efficient reasoning rl training with adaptive drafter

Reference 40

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.908296Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 7806e4fe-deeb-45ae-a94f-b608b56dc170 · outbound

This paper cites Seer: Online Context Learning for Fast Synchronous LLM Reinforcement Learning.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Seer: Online Context Learning for Fast Synchronous LLM Reinforcement Learning

Reference 41

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.780484Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation d543da68-bab7-4ee2-b074-5d912626231c · outbound

This paper cites Optimizing {RLHF} training for large language models with stage fusion.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Optimizing {RLHF} training for large language models with stage fusion

Reference 42

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.922164Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation b66f0bbe-6517-44e0-8c0b-9ed262dbc682 · outbound

This paper cites Towards efficient reward service for rlvr with request- level flexibility and batch-level constraint.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Towards efficient reward service for rlvr with request- level flexibility and batch-level constraint

Reference 43

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.958891Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 3aa6a3ef-4103-4653-b148-2dc98ca7e855 · outbound

This paper cites On the uncapacitated location problem.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning On the uncapacitated location problem

Reference 44

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.894780Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 96e454c5-a57f-4fcc-a685-22268ca5fdfb · outbound

This paper cites Linear-Programming-Based Load Balancer (LPLB).

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Linear-Programming-Based Load Balancer (LPLB)

Reference 45

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.975665Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 20129cc4-2d8f-4744-9263-5d1853b3153c · outbound

This paper cites Alibaba hpn: A data center network for large language model training.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Alibaba hpn: A data center network for large language model training

Reference 46

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.980058Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation f36e49a2-b4ff-4c96-b2c6-285808a35645 · outbound

This paper cites NVIDIA GTC: Accelerating Mixture of Experts Train- ing With Rail-Optimized InfiniBand Networking in Cru- soe Cloud.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning NVIDIA GTC: Accelerating Mixture of Experts Train- ing With Rail-Optimized InfiniBand Networking in Cru- soe Cloud

Reference 47

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.890336Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 34f396d0-c592-4030-93ba-9fda397948d6 · outbound

This paper cites GPUDirect RDMA.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning GPUDirect RDMA

Reference 48

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.898809Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 68807d47-bed7-4b07-9a64-eee85f945339 · outbound

This paper cites Optimized primitives for inter-GPU communication.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Optimized primitives for inter-GPU communication

Reference 49

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.966717Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 1b5027e2-0064-412c-9107-87ab95e2bbd8 · outbound

This paper cites DeepEP: an efficient expert-parallel communication library.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning DeepEP: an efficient expert-parallel communication library

Reference 50

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.984310Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 1dcc35f2-baff-4af2-8ad4-820eee02c33b · outbound

This paper cites {DistServe}: Disaggregating prefill and decoding for goodput-optimized large language model serving.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning {DistServe}: Disaggregating prefill and decoding for goodput-optimized large language model serving

Reference 51

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.866727Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 28703dcc-1496-442d-935c-d775a0033ecf · outbound

This paper cites Megascale- infer: Efficient mixture-of-experts model serving with disaggregated expert parallelism.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Megascale- infer: Efficient mixture-of-experts model serving with disaggregated expert parallelism

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.859126Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation b8c3b01f-b5d5-4568-a3f5-9f383d52cefd · outbound

This paper cites Disttrain: Addressing model and data heterogeneity with disaggregated train- ing for multimodal large language models.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Disttrain: Addressing model and data heterogeneity with disaggregated train- ing for multimodal large language models

Reference 53

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.876273Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 0d51a2e2-d65a-4ec5-9a50-15321224479a · outbound

This paper cites Heddle: A distributed orches- tration system for agentic rl rollout.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Heddle: A distributed orches- tration system for agentic rl rollout

Reference 54

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.769379Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation d4f8504c-7e28-43c0-9149-7e01a9ccae01 · outbound

This paper cites Bounds on multiprocessing timing anomalies.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Bounds on multiprocessing timing anomalies

Reference 55

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.871762Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation c5a7f206-30f3-45e5-bddc-ea9dca27d731 · outbound

This paper cites The SCIP Optimization Suite 9.0.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning The SCIP Optimization Suite 9.0

Reference 56

Resolution: metadata mismatch; arxiv_id, observed 2026-05-12T01:46:14.793302Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation d0431f56-14c5-43fa-89b8-55732931a853 · outbound

This paper cites Slime: An LLM post-training framework for RL Scal- ing.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Slime: An LLM post-training framework for RL Scal- ing

Reference 57

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.852858Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 2be4ef82-d728-4b9f-8c3e-65e47d51bf21 · outbound

This paper cites GPU optimized techniques for training transformer models at-scale.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning GPU optimized techniques for training transformer models at-scale

Reference 58

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.785432Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 3af36f7b-c481-4563-8992-d00f68bc5b66 · outbound

This paper cites Sglang: Efficient execution of structured lan- guage model programs.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Sglang: Efficient execution of structured lan- guage model programs

Reference 59

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.843832Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 2a85d91b-0d66-485c-861e-6ef7454e3f95 · outbound

This paper cites Open Multi-Processing.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Open Multi-Processing

Reference 60

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.839050Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation e1477a93-dc66-4203-b21a-10b453d2a826 · outbound

This paper cites NVIDIA Hopper Architecture In-Depth.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning NVIDIA Hopper Architecture In-Depth

Reference 61

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.848374Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 3f851cad-76ab-4529-89ed-ed56d4e247e6 · outbound

This paper cites an unresolved cited work.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Unresolved cited work

Reference 62

Resolution: unresolved; raw_fallback, observed 2026-05-14T06:28:58.862623Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 2b740753-657e-479f-a5a0-3935b7d11cff · outbound

This paper cites Flashattention-3: Fast and accurate atten- tion with asynchrony and low-precision.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Flashattention-3: Fast and accurate atten- tion with asynchrony and low-precision

Reference 63

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.830705Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 6832e81d-59a6-4c45-a709-76af756d5447 · outbound

This paper cites https: //github.com/NVIDIA/TransformerEngine.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning https: //github.com/NVIDIA/TransformerEngine

Reference 64

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.835254Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 11be90ab-bfc4-4ab6-8d3d-685362a04275 · outbound

This paper cites Ray: A distributed framework for emerging{AI} applications.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Ray: A distributed framework for emerging{AI} applications

Reference 65

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.855524Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation e7153983-0537-4ef1-8764-059d129ae801 · outbound

This paper cites GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

Reference 66

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.840788Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 032e3e72-ffe2-46ba-b2e0-b817a2dac756 · outbound

This paper cites Kimi k1.5: Scaling Reinforcement Learning with LLMs.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Kimi k1.5: Scaling Reinforcement Learning with LLMs

Reference 67

Resolution: verified exact; local_arxiv, observed 2026-05-12T01:46:14.801130Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation a4915de7-acd6-45b8-b6a4-9a4118b9a47a · outbound

This paper cites StreamRL: Scalable, Heterogeneous, and Elastic RL for LLMs with Disaggregated Stream Generation.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning StreamRL: Scalable, Heterogeneous, and Elastic RL for LLMs with Disaggregated Stream Generation

Reference 68

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.859743Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 80069897-0e6f-4418-aff2-9137cbe7b3c6 · outbound

This paper cites Accelerat- ing distributed {MoE} training and inference with lina.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Accelerat- ing distributed {MoE} training and inference with lina

Reference 69

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.971149Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 9a22491d-3788-4847-84dc-16b2a95b2f19 · outbound

This paper cites Tutel: Adap- tive mixture-of-experts at scale.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Tutel: Adap- tive mixture-of-experts at scale

Reference 70

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.825534Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation f41717dd-3c21-4f74-aba6-9b0211cc8618 · outbound

This paper cites HetuMoE: An Efficient Trillion-scale Mixture-of-Expert Distributed Training System.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning HetuMoE: An Efficient Trillion-scale Mixture-of-Expert Distributed Training System

Reference 71

Resolution: metadata mismatch; arxiv_id, observed 2026-05-12T01:46:14.809929Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation b4645b66-b82b-4d01-b3da-bd583d60d13a · outbound

This paper cites Megascale-moe: Large-scale communication-efficient training of mixture-of-experts models in production.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Megascale-moe: Large-scale communication-efficient training of mixture-of-experts models in production

Reference 72

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.847449Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation a00867cb-86da-4ede-8776-33825a23d0d4 · outbound

This paper cites Centauri: Enabling efficient scheduling for communication-computation overlap in large model training via communication partitioning.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Centauri: Enabling efficient scheduling for communication-computation overlap in large model training via communication partitioning

Reference 73

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.821197Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 71b27ec6-5411-444b-b686-3bb3b9c6fbfc · outbound

This paper cites {MegaScale}: Scaling large language model training to more than 10,000{GPUs}.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning {MegaScale}: Scaling large language model training to more than 10,000{GPUs}

Reference 74

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.790159Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation a999f442-c95e-4305-9e99-7f454c60e9ff · outbound

This paper cites Comet: Fine-grained computation-communication overlapping for mixture-of-experts.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Comet: Fine-grained computation-communication overlapping for mixture-of-experts

Reference 75

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.803518Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 77edfa72-05a8-4614-b0ab-497e1a131a5b · outbound

This paper cites FLUX: Fast Software-based Communication Overlap On GPUs Through Kernel Fusion.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning FLUX: Fast Software-based Communication Overlap On GPUs Through Kernel Fusion

Reference 76

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.822291Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 817bc4fb-959e-498c-8853-31be95e918e2 · outbound

This paper cites Janus: A unified distributed training framework for sparse mixture-of- experts models.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Janus: A unified distributed training framework for sparse mixture-of- experts models

Reference 77

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.816731Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 276bd38c-7d29-4d78-94fb-87cd95832341 · outbound

This paper cites Hybridflow: A flexible and efficient rlhf framework.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Hybridflow: A flexible and efficient rlhf framework

Reference 78

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.812129Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation a862306a-ed53-4214-90f2-ca0336c60cce · outbound

This paper cites Rollart: Scaling agentic rl training via disaggregated infrastructure.arXiv preprint arXiv:2512.22560.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Rollart: Scaling agentic rl training via disaggregated infrastructure.arXiv preprint arXiv:2512.22560

Reference 79

Resolution: verified exact; arxiv_id, observed 2026-05-12T01:46:14.818155Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 8166be66-f00f-4fe1-81e2-72925456e3e9 · outbound

This paper cites Fast infer- ence from transformers via speculative decoding.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Fast infer- ence from transformers via speculative decoding

Reference 80

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.949381Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Observation 10568927-0499-45ab-82f8-6c55c919ff32 · outbound

This paper cites Ef- ficient rl for llms with dynamic and online speculative decoding.

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning Ef- ficient rl for llms with dynamic and online speculative decoding

Reference 81

Resolution: verified fuzzy; raw_fallback, observed 2026-05-14T06:28:58.953806Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-07T06:34:17.273281+00:00.

Coverage vector

Reference resolution

Outbound references

Pith citing papers