INDEX

Explanations

any form of the word 'briefed'

oai_token-act-pair · gpt-3.5-turbo

references to information being shared or disseminated

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

adia

-0.64

venge

-0.62

frey

-0.61

Mother

-0.60

 granddaughter

-0.60

pite

-0.59

ãĥŁ

-0.59

eu

-0.59

bians

-0.58

birth

-0.58

POSITIVE LOGITS

 briefed

1.10

 briefings

0.98

escription

0.92

 aloud

0.87

mares

0.83

DragonMagazine

0.79

ocument

0.77

 beforehand

0.76

 briefing

0.76

icip

0.75

Activations Density 0.006%

any form of the word 'briefed'

references to information being shared or disseminated

No Comments

No Known Activations