INDEX

Explanations

phrases emphasizing a particular aspect or concept that is being highlighted

oai_token-act-pair · gpt-3.5-turbo

instances of the word "the"

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

AMA

-0.72

advertising

-0.70

bg

-0.68

ago

-0.66

sync

-0.64

restling

-0.64

Background

-0.63

OTA

-0.62

Scroll

-0.60

reb

-0.60

POSITIVE LOGITS

 slightest

1.36

 smallest

1.34

 simplest

1.22

 entirety

1.17

 hardest

1.17

 easiest

1.16

 entire

1.15

 biggest

1.11

 heaviest

1.09

 smartest

1.09

Activations Density 0.450%

phrases emphasizing a particular aspect or concept that is being highlighted

instances of the word "the"

No Comments

No Known Activations