INDEX

Explanations

names of political figures

oai_token-act-pair · gpt-3.5-turbo

highly active verbs and phrases related to gestures and actions in political or social contexts

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

pled

-0.73

 sooner

-0.69

netflix

-0.68

 equivalents

-0.66

 accordingly

-0.64

 extrap

-0.64

 ourselves

-0.64

haps

-0.63

 psycho

-0.61

DV

-0.61

POSITIVE LOGITS

 WATCHED

1.05

window

1.02

allery

0.82

OTOS

0.82

hello

0.80

 VIDEOS

0.76

ï

0.75

UTERS

0.74

 PHOTO

0.74

 photo

0.74

Activations Density 0.550%

names of political figures

highly active verbs and phrases related to gestures and actions in political or social contexts

No Comments

No Known Activations