Skip to content

Evaluation: Enhancing ASK GLIFIC performance #842

@Prajna1999

Description

@Prajna1999

Is your feature request related to a problem?
The ASK GLIFIC bot's accuracy is currently suboptimal, leading to user frustration. Improving evaluations, smart prompting, and knowledge base chunking is essential for better performance.

Describe the solution you'd like

  • Enhance evaluations for the ASK GLIFIC bot
  • Implement smart prompting techniques
  • Improve chunking of the knowledge base
  • Test improvements with the provided sample dataset
Original issue

Currently the ASK GLIFIC bot accuracy can be improved with better evals, smart prompting and good chunking of the knowledge base. Test it out with the sample dataset they have shared

Metadata

Metadata

Assignees

Labels

Type

No fields configured for Task.

Projects

Status

No status

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions