Benchmark
Webshop
A web agent evaluation benchmark released in 2022 that provides a simulated e-commerce shopping environment for assessing the performance of language models on web navigation and shopping tasks.
Source Articles
A Real Test for Self-Distillation Agentic RL: Comparing 10 Methods Against GRPOOct 15CoPE: Co-Evolving Acting and Forecasting for Language Agents, Doubling Sample EfficiencyOct 15BranchOPD: Making On-Policy Distillation More Efficient with Branch SamplingOct 12Deep Dive: CMU's Public Course on Computer Use Agents — Evolution, Evaluation, and TrainingSep 26