INDEX

Explanations

phrases starting with "it's not that" followed by a descriptor

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a contrast or rejection of a notion or idea

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

hips

-0.66

Haz

-0.63

ãĥĳ

-0.63

ãĥ¡

-0.63

Ö¼

-0.61

×Ļ×

-0.61

×Ļ

-0.61

Hew

-0.60

mi

-0.60

×ŀ

-0.56

POSITIVE LOGITS

interstitial

0.81

cher

0.75

 pesky

0.74

same

0.73

ched

0.71

angular

0.70

glomer

0.68

andestine

0.67

emic

0.67

 fateful

0.66

Activations Density 0.187%

phrases starting with "it's not that" followed by a descriptor

phrases indicating a contrast or rejection of a notion or idea

No Comments

No Known Activations