INDEX

Explanations

phrases related to shock or surprise

oai_token-act-pair · gpt-3.5-turbo

terms related to shock and shame

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

•layer 3 - surprise

No Comments

Negative Logits

APTER

-0.75

rongh

-0.64

İĭ

-0.62

Nur

-0.62

NAS

-0.62

required

-0.62

Bank

-0.59

 requisite

-0.59

 Bulg

-0.58

umper

-0.58

POSITIVE LOGITS

IELD

0.70

tty

0.68

 curtains

0.68

quished

0.68

aneously

0.67

aun

0.67

bane

0.67

 havoc

0.66

downs

0.66

 equilibrium

0.65

Activations Density 0.045%

phrases related to shock or surprise

terms related to shock and shame

No Comments

No Known Activations