INDEX

Explanations

mentions of a specific name, "Pekka"

oai_token-act-pair · gpt-3.5-turbo

the presence of the token "ek" in various contexts

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

ACTED

-0.74

appropriately

-0.69

 AMERICA

-0.66

ãĥ¼ãĥĨ

-0.62

ãĤ¤

-0.61

 diapers

-0.61

ÙĴ

-0.60

 belts

-0.60

ãĥĩãĤ£

-0.59

 Bland

-0.59

POSITIVE LOGITS

nown

1.38

kers

1.05

enstein

1.05

ansas

1.00

kies

0.95

yll

0.95

hov

0.95

anism

0.93

enzie

0.90

ker

0.90

Activations Density 0.017%

mentions of a specific name, "Pekka"

the presence of the token "ek" in various contexts

No Comments

No Known Activations