INDEX

Explanations

terms related to specific types of human identification, such as race and gender

oai_token-act-pair · gpt-3.5-turbo

suffixes indicating specific characteristics or attributes

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

ĸļ

-0.82

perty

-0.77

otation

-0.70

owitz

-0.69

HAEL

-0.66

arcity

-0.65

hitting

-0.63

 showc

-0.62

earch

-0.62

senal

-0.62

POSITIVE LOGITS

kees

0.67

 gentleman

0.65

bush

0.63

hood

0.62

ALLY

0.61

Gate

0.61

lihood

0.60

baum

0.58

 Protestant

0.58

boy

0.58

Activations Density 0.191%

terms related to specific types of human identification, such as race and gender

suffixes indicating specific characteristics or attributes

No Comments

No Known Activations