INDEX

Explanations

phrases indicating self-involvement or self-referential actions

oai_token-act-pair · gpt-3.5-turbo

phrases suggesting actions of self-identification or self-reference

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

gap

-0.70

heny

-0.70

 Feature

-0.67

grade

-0.65

lav

-0.63

jug

-0.59

ahan

-0.59

illary

-0.59

IPS

-0.59

culosis

-0.57

POSITIVE LOGITS

 pant

0.69

ortium

0.64

 hunted

0.64

æµ

0.62

isner

0.62

 ashamed

0.60

ãģĹ

0.60

 disgust

0.60

peror

0.60

 sanct

0.60

Activations Density 0.192%

phrases indicating self-involvement or self-referential actions

phrases suggesting actions of self-identification or self-reference

No Comments

No Known Activations