INDEX

Explanations

references to events or places where things are being viewed by the public

oai_token-act-pair · gpt-3.5-turbo

references to public viewing or the act of observing

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

trap

-0.82

redo

-0.78

lying

-0.70

wealth

-0.70

contract

-0.70

breaking

-0.69

nesses

-0.69

working

-0.68

ursed

-0.67

accompan

-0.67

POSITIVE LOGITS

 viewing

0.93

 "$:/

0.89

 Ratings

0.81

 obser

0.80

wat

0.79

 screens

0.77

 experien

0.74

âĸ¬

0.74

ById

0.73

ipers

0.73

Activations Density 0.012%

references to events or places where things are being viewed by the public

references to public viewing or the act of observing

No Comments

No Known Activations