INDEX

Explanations

words related to penalties or punishment

oai_token-act-pair · gpt-3.5-turbo

references to penalization or punishment

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

aeda

-0.81

through

-0.73

worth

-0.72

quickShipAvailable

-0.70

elf

-0.68

lycer

-0.66

overs

-0.65

afety

-0.64

ynthesis

-0.64

RIS

-0.64

POSITIVE LOGITS

ized

1.21

 penal

1.07

ised

0.96

izes

0.93

izing

0.92

ization

0.91

ising

0.81

ize

0.79

eties

0.78

 punished

0.76

Activations Density 0.019%

words related to penalties or punishment

references to penalization or punishment

No Comments

No Known Activations