Fixed high token usage for json type - Githubissues

Zipstack / unstract

No-code LLM Platform to launch APIs and ETL Pipelines to structure unstructured documents

https://unstract.com

GNU Affero General Public License v3.0

352 stars 29 forks source link

Fixed high token usage for json type #397

Closed Deepak-Kesavan closed 3 weeks ago

Deepak-Kesavan commented 3 weeks ago

What

Fixed high token usage for json type

Why

To stop unwanted token usage

How

Try parsing the LLM response to json before performing another LLM call to give answer in json format.

Can this PR break any existing features. If yes, please list possible items. If no, please explain why. (PS: Admins do not merge the PR without this section filled)

No. It is just enhancement

Database Migrations

Env Config

Relevant Docs

-

Related Issues or PRs

-

Dependencies Versions

-

Notes on Testing

-

Screenshots

Checklist

I have read and understood the [Contribution Guidelines]().

sonarcloud[bot] commented 3 weeks ago

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
No data about Coverage
0.0% Duplication on New Code

See analysis details on SonarCloud