INDEX

Explanations

references to conflicts involving Russia, North Korea, and terrorism, particularly in relation to military threats and adversarial 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

acebook

-0.74

icka

-0.73

NH

-0.73

NH

-0.72

wyn

-0.70

 Pension

-0.68

payer

-0.67

ournals

-0.66

 Guinness

-0.65

 Browne

-0.65

POSITIVE LOGITS

 aggress

1.23

 attacks

1.20

 onslaught

1.18

 threats

1.18

 threaten

1.14

 attack

1.14

 threat

1.12

 aggression

1.12

 menace

1.10

 invading

1.10

Activations Density 0.447%

references to conflicts involving Russia, North Korea, and terrorism, particularly in relation to military threats and adversarial actions

No Comments

No Known Activations