US 12,468,890 B1
Large language model ensemble for combinatorial retrieval in one-to-many matching tasks
Shon Mendelson, Petah Tikva (IL); Sigalit Bechler, Petah Tikva (IL); Natalie Bar Eliyahu, Petah Tikva (IL); Hadas Baumer, Petah Tikva (IL); Linoy Cohen, Petah Tikva (IL); and Tom Klein, Petah Tikva (IL)
Assigned to Intuit Inc., Mountain View, CA (US)
Filed by Intuit Inc., Mountain View, CA (US)
Filed on Jul. 11, 2025, as Appl. No. 19/266,602.
Int. Cl. G06F 40/284 (2020.01)
CPC G06F 40/284 (2020.01) 20 Claims
OG exemplary drawing
 
12. A system comprising:
a computer processor;
a data repository in communication with the computer processor and storing:
a target entry,
a token limit,
a dataset of entries which include a data subset comprising fewer entries than the dataset of entries, a plurality of candidate matching sets, a plurality of matching entries, and a selected matching set that matches the target entry,
wherein performing a many-to-one matching task on the target entry and the dataset of entries uses a number of tokens exceeding the token limit, and
a reduced number of tokens relative to the number of tokens used to perform the many-to-one matching task, but higher than the token limit,
a rule-based application executable by the computer processor on the target entry and the dataset of entries to output the data subset corresponding to the reduced number of tokens;
a sorting language model, having the token limit, executable by the computer processor on the data subset and the target entry to output the plurality of candidate matching sets and to reduce the reduced number of tokens to within the token limit; and
a matching language model executable by the computer processor on the plurality of candidate matching sets and the target entry to output, from among the plurality of candidate matching sets, the selected matching set.