INDEX

Explanations

words related to a specific proper name, "Lawrence."

oai_token-act-pair · gpt-3.5-turbo

instances of the name "Lawrence."

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

hedral

-0.92

ramid

-0.88

graded

-0.87

resso

-0.82

cffff

-0.81

raid

-0.79

izoph

-0.77

minist

-0.77

effic

-0.73

arers

-0.72

POSITIVE LOGITS

 Liver

0.96

ville

0.87

 Lawrence

0.85

 Hague

0.84

ç·

0.83

burg

0.83

Kra

0.82

 Berkeley

0.77

rence

0.77

ocity

0.77

Activations Density 0.038%

words related to a specific proper name, "Lawrence."

instances of the name "Lawrence."

No Comments

No Known Activations