INDEX

Explanations

times or instances when something is shared or distributed online

oai_token-act-pair · gpt-3.5-turbo

instances of the word "posted."

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

MODE

-0.73

rious

-0.69

SourceFile

-0.68

acter

-0.67

DERR

-0.64

Abstract

-0.63

ens

-0.62

ikan

-0.62

Â¯Â¯

-0.62

à©

-0.62

POSITIVE LOGITS

 behalf

1.37

site

1.02

eworld

1.00

eday

0.95

erous

0.93

yx

0.92

coming

0.92

sets

0.91

 Pastebin

0.91

demand

0.90

Activations Density 0.232%

times or instances when something is shared or distributed online

instances of the word "posted."

No Comments

No Known Activations