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

ATA

-0.71

 ferment

-0.71

AMS

-0.69

rika

-0.66

 Russo

-0.63

 peppers

-0.61

dyn

-0.61

 pepper

-0.61

Vo

-0.61

 knots

-0.60

POSITIVE LOGITS

ulk

1.08

 care

0.95

hess

0.87

ãĥŁ

0.79

care

0.78

 Care

0.72

perty

0.72

 misunder

0.71

aternity

0.71

issance

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.