INDEX

Explanations

hidden captions or text in a document

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Hide," indicating a focus on elements that require discretion or concealment

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

avia

-0.86

 taxpayer

-0.68

 parity

-0.66

 energ

-0.65

 productive

-0.65

 credit

-0.65

 millennial

-0.64

oint

-0.63

 lear

-0.63

 franc

-0.63

POSITIVE LOGITS

 Hide

4.40

Hide

2.74

hide

2.18

 hide

1.86

 hides

1.31

Enlarge

1.11

hid

1.09

SPONSORED

1.07

 Cloak

1.02

 Shiny

1.01

Activations Density 0.013%

hidden captions or text in a document

instances of the word "Hide," indicating a focus on elements that require discretion or concealment

No Comments

No Known Activations