INDEX

Explanations

adjectives ending in 'ness'

oai_token-act-pair · gpt-3.5-turbo

words ending in "ness."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

vice

-0.80

odium

-0.73

zynski

-0.71

USS

-0.71

dinand

-0.69

ramer

-0.68

reon

-0.66

ompl

-0.65

ellar

-0.65

chnology

-0.64

POSITIVE LOGITS

iness

1.21

ness

0.94

0.85

ï¸

0.85

fulness

0.83

lessness

0.83

iest

0.80

liness

0.77

nesses

0.77

ily

0.75

Activations Density 0.063%

adjectives ending in 'ness'

words ending in "ness."

No Comments

No Known Activations