Share your thoughts, 1 month free Claude Pro on usSee more

Preference-driven Tool Calling on MPT Context-Guided, Preference Recall

64.88P-EM

PREFINE

Updated 1mo ago

Evaluation Results

Method	Links
PREFINE 2026.04		64.88	72.76	74.75
LangMem 2026.04		64.4	64.54	67.83
Gemini-3-Flash 2026.04		62.65	72.73	74.25
PREFINE 2026.04		59.64	69.5	70.51
PREFINE 2026.04		52.41	66.74	67.85
PREFINE 2026.04		51.45	68.03	68.08
GPT-5 2026.04		51.2	62.33	64.77
RAG (Top-5) 2026.04		50.6	69.14	67.99
PREFINE 2026.04		49.88	72.65	68.71
GPT-5-mini 2026.04		47.59	65.38	66.69
PREFINE 2026.04		42.05	62.35	61.63
R1-Distill-Llama-8B 2026.04		34.94	65.12	61.03
GPT-4o-mini 2026.04		32.23	58.21	53.54
PREFINE 2026.04		32.17	59.05	54.6
Mem0 2026.04		31.93	64.59	59.79
PREFINE 2026.04		20.48	79.28	69.27
CodeGemma-7B 2026.04		18.67	38.88	38.17
R1-Distill-Qwen-7B 2026.04		13.55	33.49	31.58
Gemma-3-12B 2026.04		7.23	60.36	49.49