VESPO: Variational Sequence-Level Soft Policy Optimization for Stable Off-Policy LLM Training
Paper
โข 2602.10693 โข Published
โข 216
AI & ML interests
None defined yet.
Recent Activity
View all activity
Papers
REDSearcher: A Scalable and Cost-Efficient Framework for Long-Horizon Search Agents
VESPO: Variational Sequence-Level Soft Policy Optimization for Stable Off-Policy LLM Training
floyedย
submitted a paper to Daily Papers 12 days ago
debingzhangย
authored a paper 3 months ago
ygfrancoisย
updated a model 4 months ago
redmoe-ai-v1ย
updated a collection 4 months ago
redmoe-ai-v1ย
published a model 4 months ago
Challenges and Limitations in Fine-Tuning the Dots.OCR Model
11
#39 opened 4 months ago
by
JamesGs
redmoe-ai-v1ย
updated a collection 4 months ago
redmoe-ai-v1ย
updated a model 4 months ago
fix autoprocessor
#30 opened 5 months ago
by
kingsley01
Will it be supported by vLLM officially?
๐ ๐ 8
1
#20 opened 7 months ago
by
riverclouds
ๆฒกๆ็ฎๆไบคPR ็ป vllm ๅ transformers ๅ๏ผ
๐ 1
1
#28 opened 6 months ago
by
CHONGYOEYAT
Adding `transformers` as a library, and also mentioning the `custom_code` tag
#29 opened 6 months ago
by
ariG23498
AttributeError: module 'fitz' has no attribute 'open'
1
#21 opened 7 months ago
by
kaqmoh2025-A
Update VisionSdpaAttention to support memory efficient backend.
#27 opened 6 months ago
by
warrenwjk
yonghenglh6ย
updated a model 7 months ago
Getting these attached issues, please anyone help me
#15 opened 7 months ago
by
BipinBisht
Update modeling_dots_vision.py
๐ฅ 1
#19 opened 7 months ago
by
chenj123
Flash Attention Req When Rrunning Model
๐ 2
2
#13 opened 7 months ago
by
Chillarmo
Number of parameters
3
#12 opened 7 months ago
by
bgeneto
redmoe-ai-v1ย
updated a model 7 months ago