List your company
ResearchFormerly WizWand

Foundation Model Evaluation on AGIEval

58.1AccuracyRLVR
17.85228.30138.7549.199Jul 2, 2026
Updated 4mo ago

Evaluation results

MethodLinks
RLVRModel family=Qwen2.5VL-7B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0758.10.53437.7
C3RL w/o RefModel family=Qwen2.5VL-7B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0758.10.68219.7
C3RL (Ours)Model family=Qwen2.5VL-7B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0757.80.68815.4
RLCRModel family=Qwen2.5VL-7B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0757.50.66420.5
SCModel family=Qwen2.5VL-7B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=12026.0753.70.71120.8
BaseModel family=Qwen2.5VL-7B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.07510.54337.5
SaySelfModel family=Qwen2.5VL-7B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0750.80.74514.7
SFT+RefModel family=Qwen2.5VL-7B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0749.10.75310.5
RLVRModel family=Llama-3.2-3B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0737.30.52246.8
C3RL (Ours)Model family=Llama-3.2-3B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0737.30.6624.6
C3RL w/o RefModel family=Llama-3.2-3B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0737.20.65221.8
BaseModel family=Llama-3.2-3B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0734.80.55742.3
RLCRModel family=Llama-3.2-3B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0734.40.65811.1
SFT+RefModel family=Llama-3.2-3B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0732.10.59424
SCModel family=Llama-3.2-3B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=12026.0728.60.64435.2
SaySelfModel family=Llama-3.2-3B-Instruct, Decoding temperature=0.7, top_k=-1, top_p=1, Seed=0, 42, 20252026.0719.40.6117.2