INDEX

Explanations

phrases related to the perception or interpretation of events or information

oai_token-act-pair · gpt-3.5-turbo

phrases indicating how something is perceived or seen by others

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

atted

-0.71

eez

-0.68

iasco

-0.68

oller

-0.66

aiman

-0.65

 Secrets

-0.64

ModLoader

-0.63

ucc

-0.62

raz

-0.62

culosis

-0.62

POSITIVE LOGITS

pires

0.86

pired

0.85

 belonging

0.81

 opposed

0.80

 favoring

0.80

 expend

0.77

 closely

0.72

 synonymous

0.72

anian

0.72

criptions

0.72

Activations Density 0.088%

phrases related to the perception or interpretation of events or information

phrases indicating how something is perceived or seen by others

No Comments

No Known Activations