INDEX

Explanations

phrases related to personal relationships or interactions

oai_token-act-pair · gpt-3.5-turbo

references to various types of "pe" words, likely denoting "people" or related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

schild

-0.79

enegger

-0.72

 GOODMAN

-0.69

 Pulse

-0.67

 guiActiveUnfocused

-0.67

footed

-0.66

 Reincarn

-0.65

oldemort

-0.65

picture

-0.64

OME

-0.63

POSITIVE LOGITS

ggy

1.04

asant

1.00

eps

1.00

gging

0.99

pperc

0.99

aking

0.96

achy

0.96

eking

0.95

eper

0.93

ptic

0.92

Activations Density 0.009%

phrases related to personal relationships or interactions

references to various types of "pe" words, likely denoting "people" or related terms

No Comments

No Known Activations