INDEX

Explanations

words expressing amazement or awe

oai_token-act-pair · gpt-3.5-turbo

expressions of wonder and admiration

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Thieves

-0.70

gradient

-0.65

secondary

-0.65

FF

-0.63

containing

-0.63

dule

-0.63

 Agents

-0.63

Solid

-0.62

Short

-0.62

Gamer

-0.61

POSITIVE LOGITS

awe

1.18

htaking

0.92

urous

0.91

 aston

0.91

 incred

0.87

 amaz

0.86

ruciating

0.85

ingly

0.85

upe

0.84

 amazed

0.82

Activations Density 0.013%

words expressing amazement or awe

expressions of wonder and admiration

No Comments

No Known Activations