Hugging Face's logo

zhangzhifang
/
verl-agent

Safetensors
Model card Files
xet
 Community
verl-agent / .github /workflows

Commit History

Major Update: merge latest verl (#54)
823e1b4
unverified

Lang Feng 湛露先生 AoShen Franz Srambical BaiqingL BearBiscuit HL Blue Space tongyx361 Mantas Bakšys Mert Unsal yangwang92 Patrik Bartak Junrong Lin BearBiscuit05 Leege233 sicer wuxibin shengguangming wangfuchun-fc DuCross Qunhong Zeng Dai, Weinan runluo PPPisOK hiyouga Haibin Lin pengsun000 Hunter Zhang Changlong Xiang Long zyzshishui wangyeyeye zhaochenyang20 jrlin HollowMan6 ztonyzhao icejieke lixiaoguang12 Earl St Sauver Hongpeng Guo Wang Zhang Wang Zhang Jinn yhyang201 QQSong Changlong Yu changlyu chenhaiquan yushengsu Yusheng Su none06630663 ShareLer Xiang Long Rihong Qiu Yuhua Jiang Swtheking openhands Wei Wu commited on

[sglang] fix: add memory saver support to sglang rollout to avoid OOMs (#756)
9bbe1e6
unverified

Xiang Long jrlin commited on

Megatron checkpoint default not save hf_models, and provide model merge tool. (#780)
004f536
unverified

Blue Space commited on

ci/cd: add pylint to CI (#811)
917c852
unverified

frederrx commited on

doc: upgrade to vllm 0.8.2 (#769)
6cfac09
unverified

hiyouga commited on

[feat] Megatron checkpoint support for current Llama and Qwen models (#687)
7fe626c
unverified

Blue Space caaatch22 uygnef sharele commited on

Add GRPO CI to FSDP and Megatron simple e2e. (#711)
3e5881a
unverified

Blue Space commited on

fix: support transformers==4.50.0 (#704)
fb18b57
unverified

Lumeng Wu commited on

[ci] fix ci (#675)
d0135ce
unverified

Chi Zhang commited on

[ci] feat: move dataset.yml to another GPU (#639)
514a6fe
unverified

Chi Zhang commited on

[rollout] feat: add SGLang as rollout engine to verl (#490)
da627f4
unverified

Junrong Lin zhaochenyang20 commited on

Update e2e_vlm_geo3k.yml (#563)
b0b80cd
unverified

Chi Zhang commited on

[misc] feat: support vllm>0.7 world size 1 generation (#520)
d4d263d
unverified

Guangming Sheng commited on

recipe: PRIME algorithm (#362)
217665d
unverified

zfwww commited on

feat: support loading reward function from an external file (#452)
fb0f2a8
unverified

Lumeng Wu commited on

[CI] feat: auto cancel previous CI in the same PR (#499)
a9fd43b
unverified

tongyx361 commited on

Verl's megatron core_r0.11.0 backend successfully tested with 3D parallelism with multiple bug fixed (#495)
beb6e13
unverified

Blue Space uygnef commited on

test: Added the permission setting on the workflow (#504)
520d91d
unverified

willemjiang commited on

[misc] feat: add allgather method to dataproto (#497)
5cdf83e
unverified

Chi Zhang commited on

[ci] feat: add ci timeout (#487)
58f56ec
unverified

Chi Zhang commited on

[CI] Add e2e_ascend CI (#465)
1ac14e2
unverified

celestialli commited on

[feat] Initial support for VLMs, add Qwen2.5VL GRPO example (#386)
273a382
unverified

hiyouga commited on

megatron：Update megatron-lm to `core_r0.11.0` (#392)
0ac9704
unverified

ISEEKYAN chendong-1998 gaoziyuan BlueSpaceETO commited on

fire rollout: fix main_generation config and failed tests (#443)
f4ab309
unverified

HL commited on

rollout: FIRE sampling added. (#58)
b09c1e6
unverified

weizhech commited on

ci: Added the secrets scan action (#417)
bdee5d6
unverified

willemjiang commited on

feat: add support for ulysses sequence parallel for transformers >= 0.48 (#357)
65805c8
unverified

XFFXFF commited on

[fix] Improve the params template for generation (#351)
8e9012c
unverified

BearBiscuit commited on

distro: bump up version to v0.2.0.dev, limit vllm version (#327)
cccd999
unverified

HL commited on

[megatron] feat: support qwen2 megatron backend (#261)
5b10de9
unverified

kinman0224 commited on

fix: specify the hash version of action in scorecard.yml (#313)
e3e45fd
unverified

willemjiang commited on

Added content permissions of the workflow (#303)
b168938
unverified

willemjiang commited on

distro: make liger-kernel optional. do not rely on requirement.txt during setup (#286)
cce1ad8
unverified

HL commited on

[misc] fix install requirement (#279)
38d83a2
unverified

Guangming Sheng commited on

[testing][rollout] feat: support integration of vllm>=0.7.0 (spmd-version) (#209)
787abd3
unverified

ZSL98 zhangshulai commited on

[ci] fix: delete ckpts after saving in github runner (#272)
b6a5972
unverified

Guangming Sheng commited on

Feature/add remax support (#234)
0ab6345
unverified

znli commited on

Add stronger reward verification sandbox (#233)
b909ba2
unverified

zfwww commited on

[ckpt] feat: integrate checkpoint resume in RL ray trainer (#222)
235b904
unverified

Guangming Sheng commited on

[misc] feat: add ckpt manager in utils (#216)
be632b7
unverified

Guangming Sheng commited on

Create scorecard.yml for security scan (#215)
4ddb25b
unverified

Willem Jiang commited on

[misc] refactor: refactor config, add grpo ci (#204)
24b553f
unverified

Chi Zhang commited on

megatron: fix config error and add compute log prob interface (#186)
4f2a3ec
unverified

HL commited on

[feat, SFT] Support LigerKernel for SFT (#173)
0b2834e
unverified

Xingyao Wang commited on

[Liger-kernel] Add an option to use `_apply_liger_kernel_to_instance()` to load model (#133)
7814303
unverified

hongpeng-guo commited on

[misc] fix: gradient accumulation in seq balance and modify default vllm log level (#141)
21eabf3
unverified

Guangming Sheng commited on

[SFT] Support context parallelism for SFT (#132)
cbabcbb
unverified

Xingyao Wang Jiayi Pan openhands commited on

[BREAKING][misc] feat: change micro_batch_size to micro_batch_size_per_gpu (#136)
1b3d58f
unverified

Guangming Sheng commited on

[SFT] feat: Add LoRA support for SFT (#127)
ad105e0
unverified

Xingyao Wang Jiayi Pan openhands commited on

[ci] feat: add ci for sft trainer (#128)
38fb8f2
unverified

Guangming Sheng commited on