INDEX

Explanations

references to mental instability or the concept of being "insane."

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

fman

-0.86

arest

-0.78

adr

-0.72

çĦ

-0.72

bourg

-0.71

illes

-0.70

agues

-0.69

phies

-0.69

arers

-0.69

bis

-0.68

POSITIVE LOGITS

 asylum

0.80

 insane

0.80

 maniac

0.79

 proportions

0.72

 antics

0.71

ishly

0.70

 screaming

0.70

rid

0.67

 perm

0.67

cap

0.67

Activations Density 0.025%

references to mental instability or the concept of being "insane."

No Comments

No Known Activations