INDEX

Explanations

phrases indicating knowledge or information

oai_token-act-pair · gpt-3.5-turbo

phrases emphasizing knowledge or awareness

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

ItemTracker

-0.79

oples

-0.70

phrine

-0.67

isco

-0.67

 voucher

-0.66

issions

-0.65

pex

-0.64

 privatization

-0.64

 Gators

-0.63

otion

-0.63

POSITIVE LOGITS

lege

1.25

ledge

1.17

ledged

1.05

LED

0.82

LES

0.76

SOURCE

0.76

cut

0.74

 beforehand

0.74

 ABOUT

0.74

nothing

0.73

Activations Density 0.060%

phrases indicating knowledge or information

phrases emphasizing knowledge or awareness

No Comments

No Known Activations