INDEX

Explanations

words related to "power"

oai_token-act-pair · gpt-3.5-turbo

references to power or dominance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 tort

-0.76

las

-0.70

 Carib

-0.67

 Aval

-0.66

mal

-0.64

Lob

-0.64

 Coca

-0.63

CAD

-0.62

cas

-0.60

 coll

-0.59

POSITIVE LOGITS

ower

4.67

owers

3.17

OWER

2.02

owed

1.75

owered

1.71

owing

1.68

ows

1.52

own

1.37

ow

1.31

OW

1.30

Activations Density 0.012%

words related to "power"

references to power or dominance

No Comments

No Known Activations