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

ifty

-0.69

DEBUG

-0.67

iera

-0.67

IGHTS

-0.66

ya

-0.66

UES

-0.66

 comprises

-0.65

ords

-0.64

IPP

-0.64

 lowly

-0.63

POSITIVE LOGITS

catentry

0.70

milo

0.65

Reloaded

0.62

 Boise

0.62

 spons

0.62

 Aston

0.60

utical

0.60

 prolifer

0.60

 inst

0.60

fell

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.