Search results

yaml配置修改部分无效

Status: Open.

#243 In alibaba/ROLL;

· qilong-zhang opened on Nov 13, 2025

有关训推差异和old-log-pro

Status: Open.

#242 In alibaba/ROLL;

· gzoftju opened on Nov 13, 2025

How to use a dataset in chat format

Status: Open.

#241 In alibaba/ROLL;

· ZzzybEric opened on Nov 13, 2025

是否支持Wan2.2 5B的pipeline?

Status: Open.

#238 In alibaba/ROLL;

· gtzly opened on Nov 12, 2025

多机多卡配置

Status: Open.

#236 In alibaba/ROLL;

· Meta-YZ opened on Nov 12, 2025

Ray OOM prevention kills vLLM actors during engine initialization (collective_rpc/offload_states) unless GRPO_GROUP=1

Status: Open.

#234 In alibaba/ROLL;

· yasu-nishi opened on Nov 11, 2025

ROLL 是否可以加速reward模型（非LLM）计算与LLM GRPO的高效协同训练

Status: Open.

#230 In alibaba/ROLL;

· JoshonSmith opened on Nov 10, 2025

IPRO的评测集会开源吗

Status: Open.

#229 In alibaba/ROLL;

· RongPiKing opened on Nov 8, 2025

oom

Status: Open.

#219 In alibaba/ROLL;

· RongPiKing opened on Nov 6, 2025

Agentic training stucks for unclear reason

Status: Open.

#215 In alibaba/ROLL;

· kxfan2002 opened on Nov 5, 2025

Inquires about custom reward and tool use

Status: Open.

#213 In alibaba/ROLL;

· altria-zewei-wang opened on Nov 3, 2025

Will support Qwen-Image/Qwen-Image-Edit RL pipeline?

Status: Open.

#210 In alibaba/ROLL;

· yeungchenwa opened on Nov 2, 2025