LR / BSZ grid for MoE reference model #162

Open

Assignees

Labels

T4.2 - model exploration

Milestone

opened

Find a good configuration for learning rate and global batch size for the Qwen3 0.6BA100M model.

Metadata

Assignees

aaronkl

Labels

T4.2 - model exploration

Type

No type

Projects

WP4 - LLM building task board

Status

In Progress

Milestone

moeNo due date

Relationships

None yet

Development

No branches or pull requests