INDEX

Explanations

phrases related to legal matters and conflicts

oai_token-act-pair · gpt-3.5-turbo

references to significant historical events or figures

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

ividual

-0.78

ogun

-0.74

othal

-0.74

ancial

-0.74

uilt

-0.73

itably

-0.72

;;;;;;;;;;;;

-0.70

ecause

-0.69

irlf

-0.68

Ã©e

-0.66

POSITIVE LOGITS

CLASSIFIED

0.91

 antioxid

0.86

 behavi

0.85

Balt

0.76

NetMessage

0.74

WAYS

0.73

GoldMagikarp

0.71

 suscept

0.71

 adolesc

0.70

 reperc

0.69

Activations Density 16.626%

phrases related to legal matters and conflicts

references to significant historical events or figures

No Comments

No Known Activations