INDEX

Explanations

names or initials of people in a sports or professional context

oai_token-act-pair · gpt-3.5-turbo

references to running backs in sports contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Consent

-0.75

Magikarp

-0.73

elman

-0.69

mble

-0.68

insula

-0.67

 drift

-0.67

 plur

-0.67

ebus

-0.66

eur

-0.65

henko

-0.65

POSITIVE LOGITS

aded

0.90

RB

0.81

MJ

0.78

asted

0.76

bably

0.76

enn

0.75

anche

0.74

ading

0.73

TE

0.73

duct

0.73

Activations Density 0.011%

names or initials of people in a sports or professional context

references to running backs in sports contexts

No Comments

No Known Activations