INDEX

Explanations

proper nouns, specifically names like "Elizabeth."

oai_token-act-pair · gpt-3.5-turbo

the name "Elizabeth."

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

unal

-0.78

awaru

-0.76

senal

-0.75

²¾

-0.74

ãĥ£

-0.74

kefeller

-0.71

ursed

-0.70

yright

-0.70

acca

-0.69

packing

-0.69

POSITIVE LOGITS

 Warren

1.03

 Howell

0.92

 Elizabeth

0.88

Liu

0.86

 Holmes

0.86

 Taylor

0.83

 Olsen

0.78

 Williams

0.77

Tud

0.76

 Lynn

0.74

Activations Density 0.009%

proper nouns, specifically names like "Elizabeth."

the name "Elizabeth."

No Comments

No Known Activations