INDEX

Explanations

proper nouns related to various individuals named "Sak"

oai_token-act-pair · gpt-3.5-turbo

names or terms associated with certain individuals and entities, particularly those that are prominent or mentioned frequently

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 radiation

-0.67

 chemotherapy

-0.65

 prog

-0.64

 Commonwealth

-0.63

LEASE

-0.62

VICE

-0.62

 Scientology

-0.62

 unfocusedRange

-0.62

 Hurricanes

-0.61

 Fairfax

-0.61

POSITIVE LOGITS

amoto

1.26

urai

1.22

ule

1.10

Sak

1.09

imura

1.09

oda

1.07

ata

1.07

unin

1.05

arin

1.02

yu

1.02

Activations Density 0.006%

proper nouns related to various individuals named "Sak"

names or terms associated with certain individuals and entities, particularly those that are prominent or mentioned frequently

No Comments

No Known Activations