INDEX

Explanations

words related to criminal activities and legal actions

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "allegedly."

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

ERAL

-0.76

lite

-0.71

bern

-0.68

anim

-0.68

ilation

-0.66

egg

-0.66

Ale

-0.65

fork

-0.63

helm

-0.63

ament

-0.62

POSITIVE LOGITS

 violated

0.77

 implicated

0.76

 plotted

0.76

 involved

0.74

 infringing

0.74

 infring

0.73

 accuse

0.71

 originated

0.71

 obtained

0.71

 interfered

0.70

Activations Density 0.013%

words related to criminal activities and legal actions

occurrences of the word "allegedly."

No Comments

No Known Activations