INDEX

Explanations

locations or place names, particularly those including the word "Hills"

oai_token-act-pair · gpt-3.5-turbo

references to geographic locations, specifically the name "Hills."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

istically

-0.88

binary

-0.76

ISM

-0.74

SHIP

-0.70

thumbnails

-0.69

istic

-0.68

*/(

-0.67

emort

-0.65

cial

-0.63

xit

-0.63

POSITIVE LOGITS

borough

1.33

dale

1.27

boro

1.14

bury

0.93

ourcing

0.85

ourced

0.84

burgh

0.82

shire

0.81

enez

0.81

cliffe

0.77

Activations Density 0.025%

locations or place names, particularly those including the word "Hills"

references to geographic locations, specifically the name "Hills."

No Comments

No Known Activations