게임 정보
Metaheuristic is an application which implements (or intended to) a Turing complete machine. The main use of MH is a management of distributed tasks. Common cases of usage are:
Batch processing.
Common usage of batch processing - split data, create tasks for processing each part of data, process Tasks, aggregate results
Bulk processing of images.
Bulk processing of text.
All calcualtions are performed by distributed functions.
Metaheuristic can be deployed on-premiceses or in cloud. On-premises deployment allows to utilize internal computationa resource in more efficient way, i.e. running tasks at night. Also it helps to protect companies' intellectual property.
Metaheuristic behavior platform
Metaheuristic behavior platform is an extension for Metaheuristic for running Scenarios, evaluating and assessing quality of LLM-based APIs.
Running Scenarios
Scenario is DAG-based definition of steps for calling an external LLM-based API. For performing such calls MHBP create a prompt, collecting result, and makes additional transformation of data.
Evaluating and assessing quality
MHBP is scheduling requests to LLM and execute such requests. Collect the answers to prompt and evaluate them.
Decisions which are based on the result of evaluation.
Based on the result of evaluating and assessing of API, MHBP can create triggers, events, or call API to inform 3rd parties, if quality of tested LLM was degraded.
Batch processing.
Common usage of batch processing - split data, create tasks for processing each part of data, process Tasks, aggregate results
Bulk processing of images.
Bulk processing of text.
All calcualtions are performed by distributed functions.
Metaheuristic can be deployed on-premiceses or in cloud. On-premises deployment allows to utilize internal computationa resource in more efficient way, i.e. running tasks at night. Also it helps to protect companies' intellectual property.
Metaheuristic behavior platform
Metaheuristic behavior platform is an extension for Metaheuristic for running Scenarios, evaluating and assessing quality of LLM-based APIs.
Running Scenarios
Scenario is DAG-based definition of steps for calling an external LLM-based API. For performing such calls MHBP create a prompt, collecting result, and makes additional transformation of data.
Evaluating and assessing quality
MHBP is scheduling requests to LLM and execute such requests. Collect the answers to prompt and evaluate them.
Decisions which are based on the result of evaluation.
Based on the result of evaluating and assessing of API, MHBP can create triggers, events, or call API to inform 3rd parties, if quality of tested LLM was degraded.
