INDEX

Explanations

words related to criticism or negative comments

oai_token-act-pair · gpt-3.5-turbo

terms associated with derogatory or disparaging language

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ramid

-0.81

ħĭ

-0.80

RH

-0.75

hner

-0.71

 GOODMAN

-0.70

*/(

-0.68

HCR

-0.67

eanor

-0.66

ource

-0.66

metic

-0.66

POSITIVE LOGITS

 dispar

1.00

ately

0.93

agement

0.91

ously

0.84

aging

0.79

 Vaugh

0.78

uously

0.77

ãĥ¥

0.70

ached

0.70

 leveled

0.70

Activations Density 0.012%

words related to criticism or negative comments

terms associated with derogatory or disparaging language

No Comments

No Known Activations