ResearchBenchmarksAuto-bidding on AuctionNet (Dense)Follow382ConversionsLBM(GQPO)262.4293.45324.5355.55Mar 5, 2026Evaluation ResultsMethodMethodLinksConversionsScoreLBM(GQPO)mode=fine-tuned with GQPOmode=fine-tuned with GQPO2026.03382348LBM(P)mode=pretrainedmode=pretrained2026.03376320DT-Q2026.03371334DT2026.03364329CQL2026.03336171IQL2026.03322281BCQ2026.03321270DiffBid-Q2026.03319260DiffBid2026.03316214USCB2026.03267157