INDEX

Explanations

phrases indicating willingness for engagement or interaction

oai_token-act-pair · gpt-3.5-turbo

expressions that encourage openness and communication

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

rament

-0.77

 Templ

-0.64

ources

-0.60

 Canaver

-0.59

 Clement

-0.57

 McDonnell

-0.55

¬¼

-0.55

etheless

-0.54

oaded

-0.54

SourceFile

-0.53

POSITIVE LOGITS

inct

0.74

itsu

0.70

INE

0.66

in

0.63

inal

0.62

inem

0.62

ins

0.62

ine

0.61

inates

0.60

inyl

0.60

Activations Density 0.152%

phrases indicating willingness for engagement or interaction

expressions that encourage openness and communication

No Comments

No Known Activations