INDEX

Explanations

references to legal locations such as courthouses or courtroom-related terms

oai_token-act-pair · gpt-3.5-turbo

mentions of the term "court" in various contexts

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

 rooting

-0.67

hol

-0.67

 flashlight

-0.62

 Hungarian

-0.61

 Horus

-0.61

 culturally

-0.59

ocalypse

-0.59

 donor

-0.59

 curfew

-0.59

ulously

-0.57

POSITIVE LOGITS

tyard

1.53

thouse

1.32

cour

1.12

ser

1.05

ses

0.99

rier

0.99

 Cour

0.99

sers

0.95

icol

0.93

cil

0.92

Activations Density 0.013%

references to legal locations such as courthouses or courtroom-related terms

mentions of the term "court" in various contexts

No Comments

No Known Activations