INDEX

Explanations

the word "prime" as part of a phrase

oai_token-act-pair · gpt-3.5-turbo

references to the term "prime minister"

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

anwhile

-0.92

 Canaver

-0.82

oldemort

-0.79

iasco

-0.74

ansson

-0.73

apego

-0.71

addock

-0.70

okia

-0.70

urrencies

-0.70

afe

-0.68

POSITIVE LOGITS

 minister

0.93

 ministerial

0.93

knit

0.86

 ministers

0.81

val

0.78

iary

0.74

 office

0.74

etime

0.72

ises

0.69

mun

0.69

Activations Density 0.014%

the word "prime" as part of a phrase

references to the term "prime minister"

No Comments

No Known Activations