INDEX

Explanations

mentions of established patterns or examples to follow

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate setting or discussing precedents

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

wagen

-0.78

tein

-0.71

sung

-0.70

oma

-0.70

atto

-0.69

chel

-0.69

rill

-0.68

itate

-0.67

iew

-0.66

riv

-0.66

POSITIVE LOGITS

 precedent

1.13

 preced

0.99

DragonMagazine

0.83

setting

0.81

 precedence

0.77

 breaker

0.74

ĸļ

0.74

breakers

0.66

Setting

0.65

 paved

0.65

Activations Density 0.037%

mentions of established patterns or examples to follow

phrases that indicate setting or discussing precedents

No Comments

No Known Activations