INDEX

Explanations

phrases related to physical locations or properties

oai_token-act-pair · gpt-3.5-turbo

mentions of specific locations or premises

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

obal

-0.76

ggles

-0.71

Money

-0.69

 Twins

-0.68

olan

-0.66

avia

-0.66

UST

-0.65

gling

-0.64

ux

-0.64

toggle

-0.62

POSITIVE LOGITS

 premises

1.25

washer

0.71

hesda

0.66

 Aires

0.66

 holdings

0.66

 pubs

0.65

 grounds

0.65

hog

0.65

holders

0.64

 Malfoy

0.64

Activations Density 0.010%

phrases related to physical locations or properties

mentions of specific locations or premises

No Comments

No Known Activations