INDEX

Explanations

words related to the concept of something being wrong or incorrect

oai_token-act-pair · gpt-3.5-turbo

expressions of something being wrong or problematic

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

minist

-0.76

 Flavoring

-0.71

riber

-0.69

pole

-0.67

 corrid

-0.65

chet

-0.64

quickShipAvailable

-0.64

ribes

-0.63

cit

-0.63

è¦ļéĨĴ

-0.63

POSITIVE LOGITS

eous

0.95

headed

0.90

fully

0.87

lihood

0.70

aligned

0.68

doing

0.67

align

0.67

ful

0.67

 wing

0.67

mouth

0.66

Activations Density 0.021%

words related to the concept of something being wrong or incorrect

expressions of something being wrong or problematic

No Comments

No Known Activations