INDEX

Explanations

dates mentioned in varied formats

oai_token-act-pair · gpt-3.5-turbo

instances of dates

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

 tremend

-0.92

olulu

-0.89

estern

-0.88

raq

-0.85

HAEL

-0.84

ierrez

-0.80

unicip

-0.78

ailability

-0.77

yrus

-0.77

hell

-0.77

POSITIVE LOGITS

 rape

0.72

 dates

0.68

 Dates

0.66

 date

0.65

cake

0.64

 Dating

0.64

TBD

0.64

 Countdown

0.64

PDT

0.63

 User

0.62

Activations Density 0.014%

dates mentioned in varied formats

instances of dates

No Comments

No Known Activations