Note BaRe-Mem: Bayesian Reliability Memory for Robust and Adaptive Agent Consultation

Note The question streams: two capability regimes, 0-100% misleading advisor answers

Qwen/Qwen3-4B

Text Generation • 4B • Updated Jul 26, 2025 • 5.81M • • 716

Note Central model

Qwen/Qwen3-8B

Text Generation • 8B • Updated Jul 26, 2025 • 11.7M • • 2.06k

Note Central model

Qwen/Qwen3-14B

Text Generation • 15B • Updated Jul 26, 2025 • 2.16M • • 486

Note Central model (main text)

mistralai/Ministral-8B-Instruct-2410

8B • Updated Jul 31, 2025 • 153k • 586

Note Central model

Qwen/Qwen2.5-7B-Instruct

Text Generation • 8B • Updated Jan 12, 2025 • 9.64M • • 2.33k

Note Central model

microsoft/phi-4

Text Generation • 15B • Updated Jul 14 • 563k • • 2.31k

Note Central model (main text)

google/gemma-3-4b-it

Image-Text-to-Text • 4B • Updated Mar 21, 2025 • 1.45M • • 1.54k

Note Advisor 1 (peer_0)

microsoft/Phi-4-mini-instruct

Text Generation • 4B • Updated Dec 10, 2025 • 378k • • 844

Note Advisor 2 (peer_1)

Qwen/Qwen2.5-Coder-7B-Instruct

Text Generation • 8B • Updated Jan 12, 2025 • 2.35M • • 810

Note Advisor 3 (peer_2)

meta-llama/Llama-3.1-8B-Instruct

Text Generation • 8B • Updated Sep 25, 2024 • 6.14M • • 8.05k

Note Advisor 4 (peer_3)

deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct

Text Generation • 16B • Updated Jul 3, 2024 • 906k • 668

Note Advisor 5 (peer_4)

deepseek-ai/DeepSeek-R1-Distill-Qwen-7B

Text Generation • 8B • Updated Feb 24, 2025 • 272k • • 899

Note Advisor 6 (peer_5)