INDEX

Explanations

proper nouns or names, specifically those related to the name "Ward"

oai_token-act-pair · gpt-3.5-turbo

references to specific political wards or districts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

berus

-0.83

ivia

-0.82

kered

-0.78

alogue

-0.77

raltar

-0.77

anamo

-0.76

anchester

-0.75

ctions

-0.75

itia

-0.74

onut

-0.72

POSITIVE LOGITS

robe

1.34

 Ward

1.03

Ward

0.92

ynski

0.86

bay

0.81

 Warden

0.81

wick

0.81

lings

0.78

ling

0.77

 Passage

0.77

Activations Density 0.006%

proper nouns or names, specifically those related to the name "Ward"

references to specific political wards or districts

No Comments

No Known Activations