INDEX

Explanations

words related to war, conflict, and devastation

oai_token-act-pair · gpt-3.5-turbo

references to conflict and war-related situations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.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

ATP

-0.67

 Lunch

-0.66

ITY

-0.64

 Nickel

-0.63

 Unicode

-0.63

 Ital

-0.63

 Parenthood

-0.63

UD

-0.62

 Dinner

-0.62

 basis

-0.62

POSITIVE LOGITS

ridden

1.73

prone

1.62

induced

1.51

torn

1.47

resistant

1.47

related

1.43

laden

1.42

inducing

1.42

filled

1.35

linked

1.33

Activations Density 0.067%

words related to war, conflict, and devastation

references to conflict and war-related situations

No Comments

No Known Activations