INDEX

Explanations

proper nouns, specifically names

oai_token-act-pair · gpt-3.5-turbo

references to particular individuals, especially those named Mau and Piet

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

angers

-0.84

Interstitial

-0.81

nikov

-0.81

ãģį

-0.79

enegger

-0.77

é¾į

-0.75

é¾įå

-0.74

OUGH

-0.69

ãĤ¤ãĥĪ

-0.68

ãĤ¼

-0.68

POSITIVE LOGITS

Mau

1.11

ertodd

0.96

 Piet

0.90

aeper

0.82

iflower

0.82

catentry

0.81

pillar

0.80

vine

0.79

leted

0.76

edIn

0.76

Activations Density 0.012%

proper nouns, specifically names

references to particular individuals, especially those named Mau and Piet

No Comments

No Known Activations