INDEX

Explanations

the word "arraigned" or variations of it related to legal procedures

oai_token-act-pair · gpt-3.5-turbo

terms related to legal proceedings and arraignments

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

idge

-0.82

tz

-0.70

 Cookie

-0.65

Peb

-0.64

 Penguin

-0.64

 Penet

-0.61

iggle

-0.61

 Bunker

-0.61

OMG

-0.60

 parachute

-0.59

POSITIVE LOGITS

 arra

0.95

ignment

0.92

ysis

0.90

ctions

0.83

uations

0.78

zos

0.74

itant

0.73

hibited

0.72

abis

0.71

anooga

0.71

Activations Density 0.009%

the word "arraigned" or variations of it related to legal procedures

terms related to legal proceedings and arraignments

No Comments

No Known Activations