INDEX

Explanations

words that describe significant or impactful concepts related to morality and ethics

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

arettes

-0.86

osponsors

-0.82

unks

-0.74

ummies

-0.73

rax

-0.72

users

-0.72

parents

-0.72

UTERS

-0.72

ÃĥÃĤ

-0.72

gars

-0.71

POSITIVE LOGITS

 endeavor

1.09

 tale

1.00

 institution

0.96

 milestone

0.96

 feat

0.96

 undertaking

0.96

 topic

0.95

 avenue

0.95

 piece

0.91

 distinction

0.91

Activations Density 0.076%

words that describe significant or impactful concepts related to morality and ethics

No Comments

No Known Activations