INDEX

Explanations

political mentions and opinions, potentially of a controversial nature

oai_token-act-pair · gpt-3.5-turbo

the frequent use of the word "MORE" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ãĥ¼ãĥĨãĤ£

-0.67

ãĤ©

-0.66

Ox

-0.61

akra

-0.61

Border

-0.61

 Columb

-0.58

ature

-0.57

ãĥĥ

-0.56

Disk

-0.56

ecd

-0.55

POSITIVE LOGITS

 VIDEOS

0.96

ONSORED

0.78

osponsors

0.76

bilt

0.72

eat

0.72

ellen

0.69

DERR

0.67

 acidic

0.67

mercial

0.66

Activations Density 0.025%

political mentions and opinions, potentially of a controversial nature

the frequent use of the word "MORE" in various contexts

No Comments

No Known Activations