INDEX

Explanations

terms related to specific subjects or fields of interest

oai_token-act-pair · gpt-3.5-turbo

words related to obtaining or possession

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

 carbohyd

-0.66

 ðŁĳ

-0.65

 breathing

-0.62

 stocking

-0.60

 hemor

-0.60

 deleting

-0.60

 habit

-0.60

ebin

-0.59

 Europa

-0.59

 hist

-0.59

POSITIVE LOGITS

tain

1.48

ards

0.92

ements

0.89

kefeller

0.86

vered

0.86

erences

0.85

inen

0.84

tains

0.84

tained

0.83

ties

0.83

Activations Density 0.012%

terms related to specific subjects or fields of interest

words related to obtaining or possession

No Comments

No Known Activations