INDEX

Explanations

mentions of criminal or illegal activities

oai_token-act-pair · gpt-3.5-turbo

the word "allegedly" and related phrases indicating claims or accusations

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

lite

-0.70

bern

-0.70

ament

-0.69

eral

-0.69

ERAL

-0.67

vation

-0.65

Ale

-0.64

ger

-0.63

lake

-0.63

avascript

-0.63

POSITIVE LOGITS

 originated

0.73

 violated

0.70

 contradict

0.69

 infringing

0.68

 accuse

0.68

 infring

0.67

 implicated

0.67

 involved

0.66

 belonged

0.65

 accused

0.65

Activations Density 0.014%

mentions of criminal or illegal activities

the word "allegedly" and related phrases indicating claims or accusations

No Comments

No Known Activations