INDEX

Explanations

strings related to harmful or dangerous activities

oai_token-act-pair · gpt-3.5-turbo

terms associated with health risks and conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Daylight

-0.63

 Solitaire

-0.59

OTOS

-0.59

Ctrl

-0.56

UTF

-0.56

:]

-0.55

Priv

-0.55

 initials

-0.55

 Sirius

-0.55

nih

-0.53

POSITIVE LOGITS

roying

1.19

renched

1.08

itored

1.03

ielding

0.92

usting

0.92

ASED

0.90

iddled

0.89

ained

0.88

quartered

0.88

umping

0.87

Activations Density 0.091%

strings related to harmful or dangerous activities

terms associated with health risks and conditions

No Comments

No Known Activations