INDEX

Explanations

mentions of death tolls

oai_token-act-pair · gpt-3.5-turbo

recurring mentions of the word "toll," particularly in the context of negative impacts or consequences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Clever

-0.76

aku

-0.70

Champ

-0.63

aired

-0.62

gb

-0.61

 Holo

-0.61

ãĥİ

-0.61

asonic

-0.60

 Excellent

-0.60

prop

-0.60

POSITIVE LOGITS

 toll

4.38

 Toll

2.73

 tally

1.00

 congestion

0.94

 ransom

0.92

 freight

0.91

 hotline

0.86

 truce

0.85

 taxing

0.83

 casualty

0.82

Activations Density 0.012%

mentions of death tolls

recurring mentions of the word "toll," particularly in the context of negative impacts or consequences

No Comments

No Known Activations