INDEX

Explanations

references to various types of threats

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

urses

-0.84

ricks

-0.83

tein

-0.73

arist

-0.71

raham

-0.68

ools

-0.68

gian

-0.68

 gown

-0.67

ributes

-0.67

uties

-0.67

POSITIVE LOGITS

 posed

1.27

 threat

0.99

 threats

0.94

threat

0.87

 emanating

0.82

 Threat

0.81

 glare

0.77

lessly

0.76

xual

0.75

sov

0.75

Activations Density 0.024%

references to various types of threats

No Comments

No Known Activations