INDEX

Explanations

the word "cannot" and its variations

oai_token-act-pair · gpt-3.5-turbo

phrases expressing inability or impossibility

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

bats

-0.79

grounds

-0.70

flame

-0.68

soDeliveryDate

-0.67

 Generation

-0.67

ItemThumbnailImage

-0.65

pocket

-0.63

estate

-0.63

essen

-0.62

senal

-0.62

POSITIVE LOGITS

 afford

1.14

 necessarily

1.04

 reproduce

0.94

 tolerate

0.93

adian

0.93

 distinguish

0.88

't

0.86

 abide

0.86

 survive

0.86

 penetrate

0.85

Activations Density 0.029%

the word "cannot" and its variations

phrases expressing inability or impossibility

No Comments

No Known Activations