INDEX

Explanations

the word "row" being used in a consecutive numerical context

oai_token-act-pair · gpt-3.5-turbo

references to repeated occurrences or sequences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

merce

-0.78

urate

-0.70

iciency

-0.69

 incorporation

-0.69

omic

-0.67

anchester

-0.65

 competence

-0.65

omorphic

-0.65

 mediation

-0.64

 mathemat

-0.64

POSITIVE LOGITS

row

1.20

dy

1.16

 rows

0.88

dies

0.87

Row

0.86

span

0.76

ser

0.75

feed

0.75

Winged

0.74

cars

0.73

Activations Density 0.005%

the word "row" being used in a consecutive numerical context

references to repeated occurrences or sequences

No Comments

No Known Activations