INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

isu

-0.72

 pilgr

-0.68

SpaceEngineers

-0.68

shared

-0.67

ãĤ´ãĥ³

-0.66

 laun

-0.64

inav

-0.64

har

-0.64

isch

-0.64

coh

-0.63

POSITIVE LOGITS

eland

0.81

INESS

0.73

BAT

0.72

lies

0.72

ARDS

0.71

ENTS

0.70

---------

0.70

Batman

0.69

 Thumbnails

0.68

FORE

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.