INDEX

Explanations

highlights or emphasizes specific pieces of information in a text

oai_token-act-pair · gpt-3.5-turbo

instances of the word "highlights"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

yss

-0.83

perate

-0.82

ther

-0.74

appro

-0.74

ipel

-0.73

heed

-0.72

ldon

-0.72

agically

-0.71

gging

-0.71

uum

-0.71

POSITIVE LOGITS

 highlights

1.38

 Highlights

1.30

ãĤ¤ãĥĪ

0.94

 highlighting

0.81

 Features

0.80

shows

0.78

liga

0.77

 clips

0.76

 highlighted

0.75

 snippets

0.75

Activations Density 0.007%

highlights or emphasizes specific pieces of information in a text

instances of the word "highlights"

No Comments

No Known Activations