INDEX

Explanations

political or controversial terms and concepts

oai_token-act-pair · gpt-3.5-turbo

words related to deception or falsehoods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Wein

-0.47

 Naples

-0.46

BILITIES

-0.45

udeb

-0.44

Wak

-0.44

 CLSID

-0.43

 Santos

-0.42

SAN

-0.42

 Gunn

-0.42

¿½

-0.42

POSITIVE LOGITS

tainment

0.68

aneous

0.65

"],

0.54

etooth

0.53

")

0.53

>)

0.53

osate

0.52

itual

0.52

nown

0.52

iotic

0.50

Activations Density 0.622%

political or controversial terms and concepts

words related to deception or falsehoods

No Comments

No Known Activations