INDEX

Explanations

phrases containing the word "jailed"

oai_token-act-pair · gpt-3.5-turbo

instances of being jailed or imprisoned

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

avez

-0.67

zig

-0.63

xy

-0.63

ITS

-0.62

rx

-0.62

ciation

-0.60

Bru

-0.60

 Piercing

-0.60

ÃŁ

-0.59

thora

-0.59

POSITIVE LOGITS

 jailed

1.02

 prisoners

0.93

 inmates

0.92

 inmate

0.84

 prisoner

0.82

 imprison

0.82

 sentenced

0.81

 incarcerated

0.79

 jail

0.78

 jails

0.76

Activations Density 0.013%

phrases containing the word "jailed"

instances of being jailed or imprisoned

No Comments

No Known Activations