INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Gaul

-0.72

absor

-0.70

asses

-0.70

pri

-0.65

 runes

-0.65

nut

-0.62

 Germans

-0.62

 assimil

-0.62

nen

-0.61

 conquer

-0.61

POSITIVE LOGITS

ercise

0.78

 Patrol

0.74

 Mechanical

0.69

 Disabled

0.68

renheit

0.68

ecd

0.67

 Videos

0.66

emonium

0.65

inventoryQuantity

0.64

 Libraries

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.