INDEX

Explanations

negations indicating something is not what one might expect or assume

oai_token-act-pair · gpt-3.5-turbo

phrases indicating negation or disapproval

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

å¥

-0.72

ership

-0.69

éĥ

-0.67

éĹĺ

-0.67

PsyNetMessage

-0.67

iers

-0.66

Writer

-0.65

 Tours

-0.65

 Inventory

-0.64

luence

-0.63

POSITIVE LOGITS

 necessarily

1.31

icably

1.21

icable

1.11

 exactly

1.07

orious

1.02

epad

1.00

yet

0.98

withstanding

0.97

 entirely

0.92

eworthy

0.92

Activations Density 0.187%

negations indicating something is not what one might expect or assume

phrases indicating negation or disapproval

No Comments

No Known Activations