Abstract
The present disclosure relates to a method and a system for fine-tuning a context window for optimal utilization in generative models. The present disclosure suggests receiving an input prompt comprising a plurality of tokens and generating a baseline output based on the input prompt. Upon generating the baseline output, the present disclosure suggests evaluating at least one token of the plurality of tokens for contribution to the generated output. Subsequently, the present disclosure suggests computing rarity scores and possibility scores associated with the evaluated tokens by using a rarity matrix model and a possibility matrix model. Further, the present disclosure suggests generating a reward signal based on a combination of the baseline output, the rarity scores, and the possibility scores. Finally, the present disclosure suggests identifying and removing at least one non-contributory token based on the reward signal. As a result, the present disclosure provides enhanced resource efficiency for context utilization in generative models.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
MOHAN, SHIVAM and GADDAM, SUDHARSHAN KRISHNAKUMAR, "FINE TUNING CONTEXT WINDOW FOR OPTIMAL UTILIZATION IN GENERATIVE MODELS", Technical Disclosure Commons, (January 11, 2026)
https://www.tdcommons.org/dpubs_series/9167