INDEX

Explanations

the names "Donald Trump" and "John Trump."

oai_token-act-pair · gpt-3.5-turbo

names of prominent political figures and candidates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

--------

-0.58

sov

-0.58

eric

-0.57

 Polo

-0.54

ngth

-0.53

 onwards

-0.52

vous

-0.51

 Clubs

-0.50

cens

-0.50

chy

-0.50

POSITIVE LOGITS

 Diane

0.53

Shift

0.53

Jes

0.52

Dean

0.51

alcohol

0.51

EPA

0.51

Shield

0.50

apolis

0.50

ator

0.50

avanaugh

0.48

Activations Density 0.035%

the names "Donald Trump" and "John Trump."

names of prominent political figures and candidates

No Comments

No Known Activations