INDEX

Explanations

references to high-ranking institutions or authorities

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

vier

-0.75

arium

-0.75

 rall

-0.74

oidal

-0.72

itte

-0.71

ious

-0.71

vernment

-0.70

OPLE

-0.69

okemon

-0.68

uish

-0.66

POSITIVE LOGITS

landers

1.23

lighting

1.12

lights

1.01

 School

0.98

nesses

0.96

lander

0.93

light

0.93

school

0.91

 Definition

0.83

ness

0.83

Activations Density 0.026%

references to high-ranking institutions or authorities

No Comments

No Known Activations