INDEX

Explanations

mentions of geographical locations or regions

oai_token-act-pair · gpt-3.5-turbo

phrases indicating widespread occurrence or distribution

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

nery

-0.86

IZ

-0.66

nas

-0.66

etic

-0.60

FORE

-0.60

 ejac

-0.59

inse

-0.58

 coincidence

-0.57

nel

-0.57

 Penal

-0.57

POSITIVE LOGITS

roads

1.01

ĨĴ

0.99

urst

0.89

cedented

0.83

atform

0.82

eatures

0.80

scl

0.78

arrow

0.76

 platforms

0.75

artisan

0.75

Activations Density 0.024%

mentions of geographical locations or regions

phrases indicating widespread occurrence or distribution

No Comments

No Known Activations