INDEX

Explanations

mentions of intensifying or emphasizing words through the use of endings like '-ate' or '-ive'

oai_token-act-pair · gpt-3.5-turbo

occurrences of "ens" as a suffix or within words

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

 hitch

-0.65

zzi

-0.63

 Canaveral

-0.62

ts

-0.62

addafi

-0.61

 Able

-0.61

 McAuliffe

-0.60

 Barnett

-0.59

 Doodle

-0.59

 Tinker

-0.58

POSITIVE LOGITS

urable

1.02

terday

0.95

manship

0.92

umer

0.90

kamp

0.88

ource

0.88

uits

0.87

lander

0.86

chant

0.85

burgh

0.84

Activations Density 0.022%

mentions of intensifying or emphasizing words through the use of endings like '-ate' or '-ive'

occurrences of "ens" as a suffix or within words

No Comments

No Known Activations