INDEX

Explanations

references to places or concepts related to protection or safety

oai_token-act-pair · gpt-3.5-turbo

instances of the word "refuge" and related concepts indicating safety or shelter

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ONES

-0.75

lass

-0.75

thin

-0.70

asts

-0.69

ym

-0.69

haw

-0.69

osc

-0.68

ahon

-0.66

ike

-0.66

oths

-0.66

POSITIVE LOGITS

 refuge

0.99

 Refuge

0.93

ctuary

0.77

ashtra

0.73

natureconservancy

0.70

 Shooting

0.69

atoon

0.68

ilitation

0.66

 Deity

0.66

UGE

0.65

Activations Density 0.016%

references to places or concepts related to protection or safety

instances of the word "refuge" and related concepts indicating safety or shelter

No Comments

No Known Activations