INDEX

Explanations

proper nouns related to news articles

oai_token-act-pair · gpt-3.5-turbo

numerical values or references to measurements

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

apon

-0.71

mbuds

-0.65

apons

-0.63

 Tail

-0.62

 Magical

-0.60

boro

-0.60

emies

-0.59

inyl

-0.57

geon

-0.56

 âĸĪ

-0.55

POSITIVE LOGITS

Prev

0.73

Sensor

0.69

BBC

0.65

Channel

0.62

venge

0.62

elsen

0.62

 landfall

0.59

 Adinida

0.59

Bus

0.58

GMT

0.58

Activations Density 0.247%

proper nouns related to news articles

numerical values or references to measurements

No Comments

No Known Activations