INDEX

Explanations

the most recent or up-to-date information or events

oai_token-act-pair · gpt-3.5-turbo

instances of the word "latest" in context with reports or developments

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

avery

-0.80

erved

-0.75

hovah

-0.74

velt

-0.74

oak

-0.73

par

-0.72

aldehyde

-0.71

dinand

-0.70

krit

-0.68

ships

-0.68

POSITIVE LOGITS

 incarnation

1.25

 iteration

1.20

 installment

1.16

 edition

1.08

 developments

0.92

 arrivals

0.91

 round

0.90

 episode

0.89

 batch

0.88

 update

0.87

Activations Density 0.016%

the most recent or up-to-date information or events

instances of the word "latest" in context with reports or developments

No Comments

No Known Activations