INDEX

Explanations

mentions or references to being prohibited or excluded from something

oai_token-act-pair · gpt-3.5-turbo

words related to restrictions or prohibitions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 downt

-0.73

Writer

-0.71

 flyer

-0.67

 joints

-0.65

ativity

-0.65

sie

-0.65

 messenger

-0.65

 drums

-0.64

 PROG

-0.62

RIC

-0.62

POSITIVE LOGITS

arov

2.30

 barred

2.02

allow

1.96

allows

1.69

allowed

1.53

 disqual

1.38

imming

1.30

imposed

1.29

 disqualified

1.23

imov

1.05

Activations Density 0.039%

mentions or references to being prohibited or excluded from something

words related to restrictions or prohibitions

No Comments

No Known Activations