INDEX

Explanations

mentions of the word "parents."

oai_token-act-pair · gpt-3.5-turbo

references to parents

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Flavoring

-0.95

ocl

-0.72

atility

-0.70

CONT

-0.69

HRC

-0.68

JFK

-0.66

ACTED

-0.65

 Crimes

-0.64

psey

-0.63

jriwal

-0.61

POSITIVE LOGITS

hesis

1.22

hetical

1.20

heses

1.13

parents

1.10

hetically

0.98

Parents

0.93

hood

0.87

 Parents

0.85

uci

0.83

father

0.83

Activations Density 0.023%

mentions of the word "parents."

references to parents

No Comments

No Known Activations