INDEX

Explanations

phrases related to negative emotions such as hate

oai_token-act-pair · gpt-3.5-turbo

expressions of strong negative sentiments, particularly hate

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

estern

-0.87

 Bella

-0.69

 Polly

-0.68

Reviewer

-0.67

Sah

-0.67

 Cantor

-0.65

Ã«

-0.65

 Kirst

-0.65

 Elizabeth

-0.64

foreseen

-0.64

POSITIVE LOGITS

oon

0.72

ovie

0.67

 encount

0.67

 indu

0.66

 conduc

0.66

SpaceEngineers

0.63

 Tayyip

0.62

Ń

0.62

 destro

0.62

erial

0.61

Activations Density 0.000%

phrases related to negative emotions such as hate

expressions of strong negative sentiments, particularly hate

No Comments

No Known Activations