INDEX

Explanations

phrases related to specific terms and technical jargon

oai_token-act-pair · gpt-3.5-turbo

references to the term "guilty" and its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

hips

-0.85

marked

-0.82

umbledore

-0.80

weeney

-0.79

ikarp

-0.78

killer

-0.74

rider

-0.74

 icing

-0.72

fully

-0.70

battle

-0.68

POSITIVE LOGITS

ibi

0.94

ilty

0.79

 Netanyahu

0.78

acl

0.77

ÄŁ

0.74

ose

0.74

ffer

0.73

vernment

0.70

 Mesh

0.68

 Hasan

0.66

Activations Density 0.007%

phrases related to specific terms and technical jargon

references to the term "guilty" and its variations

No Comments

No Known Activations