INDEX

Explanations

references to Afghanistan and Afghan individuals

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

creen

-1.12

tower

-0.82

ometimes

-0.80

hare

-0.71

nces

-0.70

igne

-0.69

FUL

-0.69

ayers

-0.68

retty

-0.67

 Harden

-0.67

POSITIVE LOGITS

istani

1.36

istan

1.33

 Afghan

0.97

 Taliban

0.96

 Afgh

0.90

ghan

0.88

 Afghans

0.85

 civilians

0.80

 Kurd

0.80

rug

0.79

Activations Density 0.007%

references to Afghanistan and Afghan individuals

No Comments

No Known Activations