INDEX

Explanations

phrases related to specific names and possibly their associations

oai_token-act-pair · gpt-3.5-turbo

mentions of prominent media entities and figures, particularly in journalism and filmmaking

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

olulu

-0.93

ĪĴ

-0.89

antha

-0.80

culosis

-0.79

haul

-0.78

sym

-0.75

ĵĺ

-0.75

nea

-0.74

ĻĤ

-0.72

izons

-0.71

POSITIVE LOGITS

 Spiegel

1.09

atial

0.80

 Spur

0.77

Spo

0.74

 Sass

0.73

 Pool

0.73

SPI

0.73

Spr

0.71

inelli

0.71

ONSORED

0.71

Activations Density 0.021%

phrases related to specific names and possibly their associations

mentions of prominent media entities and figures, particularly in journalism and filmmaking

No Comments

No Known Activations