INDEX

Explanations

instances of people or objects physically turning around

oai_token-act-pair · gpt-3.5-turbo

instances of turning or rotating actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rought

-0.65

ites

-0.60

 addiction

-0.58

 implants

-0.58

ulla

-0.57

erness

-0.55

çļ

-0.55

cliffe

-0.54

lements

-0.54

inks

-0.54

POSITIVE LOGITS

 sideways

0.82

lda

0.72

emort

0.72

 abruptly

0.71

 startled

0.69

 scrolling

0.69

 GOODMAN

0.68

peed

0.66

undo

0.66

0.65

Activations Density 0.156%

instances of people or objects physically turning around

instances of turning or rotating actions

No Comments

No Known Activations