INDEX

Explanations

phrases related to negative remarks or insults

oai_token-act-pair · gpt-3.5-turbo

terms associated with disparaging comments and references to salvage

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

ħĭ

-1.05

 Shack

-0.88

è£ħ

-0.72

 ingred

-0.70

 mileage

-0.69

 tremend

-0.69

*/(

-0.67

ļéĨĴ

-0.65

HCR

-0.65

BRE

-0.65

POSITIVE LOGITS

aging

1.27

agement

1.17

aged

1.17

atory

1.09

ages

1.05

age

1.01

assing

0.99

ific

0.96

agall

0.96

aband

0.95

Activations Density 0.026%

phrases related to negative remarks or insults

terms associated with disparaging comments and references to salvage

No Comments

No Known Activations