INDEX

Explanations

words related to strong negative emotions like "hated"

oai_token-act-pair · gpt-3.5-turbo

references to the concept of hatred towards individuals or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

arta

-0.88

ItemImage

-0.83

SPONSORED

-0.83

OGR

-0.81

aer

-0.80

gran

-0.79

ramid

-0.79

aqu

-0.78

authent

-0.78

tein

-0.76

POSITIVE LOGITS

 hated

1.34

 hates

0.98

 hate

0.91

 despised

0.83

 prick

0.74

 disliked

0.73

 spit

0.71

 hating

0.70

 nails

0.69

 unfairly

0.69

Activations Density 0.011%

words related to strong negative emotions like "hated"

references to the concept of hatred towards individuals or entities

No Comments

No Known Activations