Skip to content

Perf: Explain text vs token perf on TRTLLM #182

@arekay-nv

Description

@arekay-nv

There is a perf regression between the using tokens versus text with TRTLLM.

Metadata

Metadata

Assignees

Labels

area: metricsEvent recorder, metrics reporter, reportingpriority: P1High — must address this cycletype: performancePerformance regression or improvement

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions