INDEX

Explanations

mentions of individuals or entities being actively involved in different situations or activities

oai_token-act-pair · gpt-3.5-turbo

references to participants or entities engaged in a situation or event

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

Jet

-0.68

vironment

-0.66

ppel

-0.65

ander

-0.64

mares

-0.64

efully

-0.64

haar

-0.64

ipe

-0.62

peat

-0.61

\\\\\\\\

-0.60

POSITIVE LOGITS

ioned

0.80

 involved

0.73

involved

0.73

olved

0.69

 implicated

0.67

enza

0.64

reon

0.64

cius

0.62

 investigating

0.61

arya

0.60

Activations Density 0.022%

mentions of individuals or entities being actively involved in different situations or activities

references to participants or entities engaged in a situation or event

No Comments

No Known Activations