INDEX

Explanations

inappropriate language or offensive terms

oai_token-act-pair · gpt-3.5-turbo

intense emotional expressions, particularly those involving strong profanity

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

Ã¨re

-0.76

Msg

-0.76

Folder

-0.74

anu

-0.73

BIL

-0.72

laus

-0.71

lication

-0.70

NetMessage

-0.70

inel

-0.69

ilus

-0.69

POSITIVE LOGITS

 kidding

0.95

 fucking

0.89

 fuck

0.82

 asshole

0.81

 bast

0.81

 goddamn

0.80

 shit

0.78

 idiots

0.75

 dick

0.75

 fuckin

0.75

Activations Density 0.027%

inappropriate language or offensive terms

intense emotional expressions, particularly those involving strong profanity

No Comments

No Known Activations