List your company
ResearchFormerly WizWand

Multimodal Conversation on MMRole (ID)

95.3LLM-as-a-Judge ScoreDr.GRPO (Latent Action)
71.977.97584.0590.125Jan 12, 2026
Updated 8mo ago

Evaluation results

MethodLinks
Dr.GRPO (Latent Action)Backbone=Qwen2.5-VL-3B-Instruct, Training Method=Dr.GRPO, Action Space=Latent Action2026.0195.3
GRPO (Latent Action)Backbone=Qwen2.5-VL-3B-Instruct, Training Method=GRPO, Action Space=Latent Action2026.0194.9
DAPO (Latent Action)Backbone=Qwen2.5-VL-3B-Instruct, Training Method=DAPO, Action Space=Latent Action2026.0194.1
BNPO (Latent Action)Backbone=Qwen2.5-VL-3B-Instruct, Training Method=BNPO, Action Space=Latent Action2026.0194
GRPO (Latent Action)Backbone=Qwen2.5-VL-7B-Instruct, Training Method=GRPO, Action Space=Latent Action2026.0192
DAPO (Latent Action)Backbone=Qwen2.5-VL-7B-Instruct, Training Method=DAPO, Action Space=Latent Action2026.0192
Dr.GRPO (Latent Action)Backbone=Qwen2.5-VL-7B-Instruct, Training Method=Dr.GRPO, Action Space=Latent Action2026.0191.6
BNPO (Latent Action)Backbone=Qwen2.5-VL-7B-Instruct, Training Method=BNPO, Action Space=Latent Action2026.0191.6
BNPO (Token)Backbone=Qwen2.5-VL-7B-Instruct, Training Method=BNPO, Action Space=Token2026.0189.4
GRPO (Token)Backbone=Qwen2.5-VL-7B-Instruct, Training Method=GRPO, Action Space=Token2026.0189.2
Dr.GRPO (Token)Backbone=Qwen2.5-VL-7B-Instruct, Training Method=Dr.GRPO, Action Space=Token2026.0189.2
DAPO (Token)Backbone=Qwen2.5-VL-7B-Instruct, Training Method=DAPO, Action Space=Token2026.0189.2
SFTBackbone=Qwen2.5-VL-7B-Instruct, Training Method=SFT, Action Space=Token2026.0188.5
Dr.GRPO (Token)Backbone=Qwen2.5-VL-3B-Instruct, Training Method=Dr.GRPO, Action Space=Token2026.0186.7
BNPO (Token)Backbone=Qwen2.5-VL-3B-Instruct, Training Method=BNPO, Action Space=Token2026.0186
DAPO (Token)Backbone=Qwen2.5-VL-3B-Instruct, Training Method=DAPO, Action Space=Token2026.0185.6
SFTBackbone=Qwen2.5-VL-3B-Instruct, Training Method=SFT, Action Space=Token2026.0184.3
PromptBackbone=Qwen2.5-VL-7B-Instruct, Training Method=Prompt, Action Space=Token2026.0183.9
GRPO (Token)Backbone=Qwen2.5-VL-3B-Instruct, Training Method=GRPO, Action Space=Token2026.0183.8
PromptBackbone=Qwen2.5-VL-3B-Instruct, Training Method=Prompt, Action Space=Token2026.0172.8