Yash Rajeshbhai Parikh

Code, Prompts, and Cached Model Responses for Seed-Alert Laundering-Ring Resolution With Tool-Using Language-Model Agents

Code, prompts, results, and cached model responses for the article "A Capability Threshold for Large Language Model Agents in Anti-Money Laundering: A Controlled Evaluation of Twenty-One Models" (Y. R. Parikh and E. K. Saraiya, submitted to IEEE Access). An investigation starts from one alerted laundering transaction a...

Replication package for "When Is Smaller Cheaper? Break-Even Tests for Large Language Model Cascades on a Metered Cloud Inference Service

Collected model traces, analysis harness, collection scripts and article source behind every number, table and figure in the article. Running reproduce.sh regenerates them offline. Version 2 accompanies the Journal of Cloud Computing submission and corrects the savings ceiling (Proposition 2) and the answer grader; ver...

Replication package for "When Smaller Is Not Cheaper: The Verbosity Inversion in Large Language Model Cascades"