INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ãĥĥãĥī

-0.76

DERR

-0.75

ffen

-0.74

KC

-0.71

IGHTS

-0.70

TN

-0.68

RESULTS

-0.68

bands

-0.67

LIN

-0.66

Beck

-0.66

POSITIVE LOGITS

hai

0.76

 radius

0.72

ra

0.72

rum

0.70

raint

0.67

eday

0.65

umbledore

0.63

inia

0.62

adesh

0.62

abad

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.