INDEX

Explanations

names of places or characters, especially "Castle"

oai_token-act-pair · gpt-3.5-turbo

mentions of "Castle" and related terms

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

Allah

-0.76

 Malays

-0.74

 doct

-0.70

tips

-0.69

Austral

-0.68

PT

-0.67

 prof

-0.67

 cleans

-0.66

 recip

-0.66

 radi

-0.65

POSITIVE LOGITS

 Castle

3.97

 castle

2.46

 castles

2.22

castle

1.88

 Cathedral

1.46

 Mansion

1.29

 Abbey

1.29

Cast

1.29

 Tower

1.24

 Citadel

1.24

Activations Density 0.011%

names of places or characters, especially "Castle"

mentions of "Castle" and related terms

No Comments

No Known Activations