INDEX

Explanations

mentions of someone being alive

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "alive."

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

pled

-0.71

ple

-0.71

RECT

-0.69

ples

-0.63

agree

-0.61

ĳ

-0.61

 Aerospace

-0.60

arov

-0.60

CHAT

-0.60

ãĥĩ

-0.60

POSITIVE LOGITS

abouts

0.97

lihood

0.92

 alive

0.87

nces

0.81

lier

0.78

 Alive

0.76

mares

0.74

hart

0.72

beat

0.69

guards

0.69

Activations Density 0.016%

mentions of someone being alive

occurrences of the word "alive."

No Comments

No Known Activations