8000 Issues · volcengine/verl · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

Issues: volcengine/verl

Beta
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

534 Open 333 Closed
534 Open 333 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Error when converting mcore to hf after RL training
#2024 opened Jun 14, 2025 by alanayu
Why does GRPO training using Lora gpu memory not decrease?
#2022 opened Jun 14, 2025 by vllbc
converter_hf_to_mcore error after pull code
#2020 opened Jun 14, 2025 by llp1992
1
Dependencies for B200
#2015 opened Jun 14, 2025 by yujianll
1
Does AsyncvLLMServer Not Yet Support VLM?
#2013 opened Jun 13, 2025 by mantle2048
training GRPO using NPU, exit with error
#2012 opened Jun 13, 2025 by xlinsz
TypeError in LoRA parameter collection due to incorrect device type checking
#2011 opened Jun 13, 2025 by Eisenhower
[performance] verl 0.4.0 slower than 0.3.0 in actor fwd/bwd part
#2009 opened Jun 13, 2025 by UnpureRationalist
3
Is sft in verl still a toy demo or is it now production-ready and widely usable?
#2005 opened Jun 13, 2025 by VoiceBeer
megatron + vllm async: AttributeError: 'vLLMRollout' object has no attribute 'execute_method'
#2001 opened Jun 12, 2025 by leoleoasd
9
Unexpected meta_info field in async_generate output, but no such field in the engine implementation
#1997 opened Jun 12, 2025 by Ericva
Compatibility with vLLM 0.9.1 for CUDA 12.8
#1993 opened Jun 12, 2025 by randydl
1
[RFC] To resolve CPU memory peak in checkpointing process: migration to megatron dist_ckpt?
#1992 opened Jun 12, 2025 by ETOgaosion
[BUG?]loss_mask?
#1989 opened Jun 12, 2025 by jzstudent
3
When the GRPO model finishes training, the program reports an error：A worker died or was killed while executing a task by an unexpected system error.
#1986 opened Jun 12, 2025 by FlAmmmmING
[ray_trainer] function **apply_kl_penalty** bug
#1984 opened Jun 12, 2025 by zhaojingqian
2
[performance] vllm async cost more gpu memory than sync mode
#1982 opened Jun 12, 2025 by Jinghui-Wang
[sft] issue: Assigning CUDA Gradient to CPU Tensor with SFT-LoRA and Offload Options
#1978 opened Jun 12, 2025 by thelongestusernameofall
Issues with VLM GRPO with verl==0.4.0 due to qwen_vl_utils
#1976 opened Jun 11, 2025 by wizeng23
Wrong doc link in release message
#1975 opened Jun 11, 2025 by leoleoasd
SGLang model weight offload
#1973 opened Jun 11, 2025 by UnpureRationalist
pr #1629 breaks autograd when entropy_coeff != 0
#1970 opened Jun 11, 2025 by linxxx3
2
training curve inconsistent with EasyR1
#1967 opened Jun 11, 2025 by Elliott011
Loss value discrepancy between SP and non-SP modes with balance_dp_token=True
#1963 opened Jun 11, 2025 by wenxueru
4
Parallelize/Async-ify/... custom_reward_function rollout
#1962 opened Jun 11, 2025 by jinulee-v
ProTip! no:milestone will show everything without a milestone.
0