INDEX

Explanations

mentions of physical illness or negative health conditions

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "sick"

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

 unlaw

-0.75

 unden

-0.67

 compr

-0.67

ETHOD

-0.65

 Unch

-0.64

inational

-0.64

merce

-0.63

anship

-0.62

AUT

-0.62

adr

-0.61

POSITIVE LOGITS

ening

1.20

bay

1.14

ened

1.06

nesses

0.94

estro

0.87

le

0.87

ly

0.86

est

0.85

ert

0.85

ness

0.82

Activations Density 0.016%

mentions of physical illness or negative health conditions

occurrences of the word "sick"

No Comments

No Known Activations