Search results

多轮数据格式

bugSomething isn't workingSomething isn't working

Status: Open.

#10966 In hiyouga/EasyR1;

· AKU-hub opened on Apr 13, 2026

[Dual Cards] Error while attaching to shared memory segment /dev/shm/nccl-Ќ (size 0), error: No such file or directory (2)

Status: Open.

#505 In hiyouga/EasyR1;

· Vimos opened on Sep 18, 2025

why you guys chose to remove max_num_seqs for vllms and remove scheduler

Status: Open.

#489 In hiyouga/EasyR1;

· menglin0320 opened on Sep 5, 2025

[Feature Request] support vLLM AsyncLLM for rollout

Status: Open.

#488 In hiyouga/EasyR1;

· ZongqiangZhang opened on Sep 4, 2025

Training time not decrease linearly during multi-node training.

Status: Open.

#483 In hiyouga/EasyR1;

· Kun-Xiang opened on Aug 30, 2025

Support Intern-s1 ?

Status: Open.

#477 In hiyouga/EasyR1;

· manglu097 opened on Aug 22, 2025

Is it insufficient memory or a memory leak?

Status: Open.

#469 In hiyouga/EasyR1;

· cloud-zZ opened on Aug 19, 2025

reinforce++ 训练实践

Status: Open.

#463 In hiyouga/EasyR1;

· manglu097 opened on Aug 14, 2025

How to add reward model to compute reward signal?

Status: Open.

#452 In hiyouga/EasyR1;

· Kun-Xiang opened on Aug 4, 2025

Does GRPO training of VLM support using a Hugging Face model as the rollout-inference model?

Status: Open.

#446 In hiyouga/EasyR1;

· shengyuwoo opened on Jul 31, 2025

Training Memory Leak at various steps

Status: Open.

#444 In hiyouga/EasyR1;

· max-vkl opened on Jul 30, 2025

PG loss does not converge

Status: Open.

#443 In hiyouga/EasyR1;

· tdano1 opened on Jul 29, 2025