INDEX

Explanations

phrases related to societal norms, beliefs, professions, citizenship, and roles

oai_token-act-pair · gpt-3.5-turbo

terms related to social roles and identities

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

forestation

-0.64

 unbeliev

-0.61

 insanity

-0.60

ãĥį

-0.60

plane

-0.59

osphere

-0.59

osi

-0.58

 Sense

-0.57

 vacancy

-0.56

Availability

-0.56

POSITIVE LOGITS

hips

1.20

mith

1.12

 themselves

1.12

hip

1.06

pring

0.95

paces

0.91

 unto

0.91

cale

0.89

ometimes

0.88

cript

0.87

Activations Density 0.254%

phrases related to societal norms, beliefs, professions, citizenship, and roles

terms related to social roles and identities

No Comments

No Known Activations