INDEX

Explanations

instances where something is available for action or consumption

oai_token-act-pair · gpt-3.5-turbo

phrases indicating availability or options for downloading, purchasing, or booking something

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

orio

-0.81

rums

-0.77

rious

-0.76

ymes

-0.74

nik

-0.74

ijn

-0.73

orman

-0.72

 Tycoon

-0.70

stem

-0.69

nect

-0.68

POSITIVE LOGITS

 download

1.15

 purchase

0.88

 downloads

0.82

download

0.78

 Purchase

0.78

 downloadable

0.73

 Download

0.70

 Downloads

0.70

 Kindle

0.65

 downloading

0.65

Activations Density 0.187%

instances where something is available for action or consumption

phrases indicating availability or options for downloading, purchasing, or booking something

No Comments

No Known Activations