INDEX

Explanations

words related to subverting or subversions

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

utical

-0.82

llah

-0.78

accompan

-0.77

ptive

-0.73

andro

-0.70

Interstitial

-0.68

çĦ

-0.66

senal

-0.65

ecause

-0.61

Bei

-0.61

POSITIVE LOGITS

ibility

1.07

ible

0.94

ophobia

0.88

icals

0.84

verted

0.82

ibly

0.82

urous

0.81

ibles

0.81

version

0.81

igo

0.80

Activations Density 0.029%

words related to subverting or subversions

No Comments

No Known Activations