INDEX

Explanations

references to supernatural or monstrous entities and their actions

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

olicy

-0.79

 Playoff

-0.78

raltar

-0.74

ourt

-0.71

 Commons

-0.70

 liberal

-0.69

 mainline

-0.69

FSA

-0.69

ffen

-0.69

izont

-0.68

POSITIVE LOGITS

 humanoid

1.22

 haunting

1.17

 mysteriously

1.17

 Cthulhu

1.13

 abduct

1.08

 mysterious

1.07

 haun

1.05

 hypnot

1.04

Cth

1.03

 haunt

1.03

Activations Density 0.251%

references to supernatural or monstrous entities and their actions

No Comments

No Known Activations