INDEX

Explanations

instances where something stops momentarily or takes a break

oai_token-act-pair · gpt-3.5-turbo

instances of the word "pause" or related forms of it

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

IFrame

Link

Negative Logits

mis

-0.75

MET

-0.71

posted

-0.70

ipped

-0.68

MAT

-0.67

orned

-0.66

alez

-0.65

olog

-0.64

opter

-0.63

lar

-0.62

POSITIVE LOGITS

 pause

1.04

olicy

0.94

 pauses

0.90

pause

0.77

 paused

0.77

gap

0.73

 puberty

0.72

 periods

0.70

 seiz

0.69

cffffcc

0.67

Activations Density 0.020%

instances where something stops momentarily or takes a break

instances of the word "pause" or related forms of it

No Known Activations