AVLM Fine-tuning Dashboard

{{ tooltip.run?.Hardware_clean }}

Backend: {{ tooltip.run?.Backend }} Method: {{ tooltip.run?.['Training Method'] }} Sharding: {{ tooltip.run?.Sharding }} AC: {{ tooltip.run?.['Activation Checkpointing'] }} TP: {{ tooltip.run?.['Tensor parallel size'] }} EP: {{ tooltip.run?.['Expert parallel size'] }}

{{ m }}	{{ fmtMetric(tooltip.run?.[m], m) }}

Cost per million tokens (relative)	{{ fmtCost(tooltip.run?.['Cost per million tokens (relative)']) }}

{{ v.label }}

Filters

Model

{{ activeFilters.Model.length }} {{ filterLabel('Model') }} ▼

All

{{ opt }}

Hardware

{{ opt }} ({{ hwCount(opt) }})

{{ FILTER_LABELS[col] }}

{{ activeFilters[col].length }} {{ filterLabel(col) }} ▼

All

{{ opt }}

Best Observed Configurations

Selected metric: {{ currentView?.label }}

Chart

Table

Recommended configurations (CSV)

All filtered runs (CSV)

No data — adjust filters

{{ tableSelected.size }} of {{ recommendedData.length }} recommended rows selected

	{{ tableHeader(col) }}