INDEX

Explanations

phrases indicating a significant or crucial point in a discussion or argument

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a point of focus or emphasis within a discussion

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

visors

-0.70

itialized

-0.62

omore

-0.58

 hygiene

-0.56

Doctors

-0.56

regate

-0.55

ONSORED

-0.55

 Bund

-0.54

 Eclipse

-0.54

 Trace

-0.54

POSITIVE LOGITS

abouts

1.57

tical

1.35

tics

1.31

tic

1.20

here

0.85

fires

0.79

 guiActiveUn

0.76

with

0.75

upon

0.73

adays

0.70

Activations Density 0.045%

phrases indicating a significant or crucial point in a discussion or argument

phrases indicating a point of focus or emphasis within a discussion

No Comments

No Known Activations