INDEX

Explanations

information related to armed conflicts, wars, and violent incidents

oai_token-act-pair · gpt-3.5-turbo

phrases referencing significant events that result in casualties or loss of life

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

isSpecialOrderable

-0.79

 learns

-0.72

 Dialogue

-0.71

 affirmative

-0.68

 Fair

-0.67

 Pros

-0.66

 Plays

-0.65

 representations

-0.64

Opt

-0.64

 Recomm

-0.63

POSITIVE LOGITS

 ravaged

1.30

 devastated

1.28

 engulfed

1.18

 raged

1.17

 engulf

1.16

 killed

1.14

 rocked

1.09

 crippled

1.09

 devast

1.06

 displaced

1.05

Activations Density 0.167%

information related to armed conflicts, wars, and violent incidents

phrases referencing significant events that result in casualties or loss of life

No Comments

No Known Activations