INDEX

Explanations

phrases containing the word "row" preceded by various prefixes

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals or entities, particularly names

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

£ı

-0.87

 nomine

-0.86

 metic

-0.84

urated

-0.79

 mosqu

-0.79

ãĥ¤

-0.75

 Gujar

-0.70

 incent

-0.68

 improbable

-0.68

 Haram

-0.67

POSITIVE LOGITS

restling

1.15

icz

1.07

row

1.07

atts

1.01

ski

1.00

dy

0.96

orld

0.92

een

0.90

zee

0.84

agy

0.84

Activations Density 0.008%

phrases containing the word "row" preceded by various prefixes

references to specific individuals or entities, particularly names

No Comments

No Known Activations