INDEX

Explanations

phrases related to media broadcasts

oai_token-act-pair · gpt-3.5-turbo

references to broadcasting or media presence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Cla

-0.73

enance

-0.71

 Saur

-0.68

 Principles

-0.67

 Burgess

-0.66

 Barcl

-0.65

 Managing

-0.65

 Mons

-0.64

 Rein

-0.64

 McKin

-0.63

POSITIVE LOGITS

waves

1.34

bnb

1.22

plane

1.14

planes

0.97

 conditioning

0.92

bender

0.90

raid

0.89

condition

0.82

mail

0.81

 turbulence

0.80

Activations Density 0.019%

phrases related to media broadcasts

references to broadcasting or media presence

No Comments

No Known Activations