INDEX

Explanations

references to planets, especially with high activation for "planet Earth"

oai_token-act-pair · gpt-3.5-turbo

references to celestial bodies, specifically planets

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

headers

-0.85

Repe

-0.76

 Marketable

-0.74

WER

-0.66

RFC

-0.62

NetMessage

-0.62

ICLE

-0.61

IME

-0.61

 disclaim

-0.61

EStreamFrame

-0.60

POSITIVE LOGITS

arium

1.56

oid

1.27

 Earth

1.18

 Venus

1.15

 orbiting

1.14

oids

1.08

 habitable

1.07

ouched

1.07

 Ceres

1.04

 earth

1.03

Activations Density 0.047%

references to planets, especially with high activation for "planet Earth"

references to celestial bodies, specifically planets

No Comments

No Known Activations