INDEX

Explanations

locations or places

oai_token-act-pair · gpt-3.5-turbo

geographical locations and city names

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

vana

-0.70

thood

-0.67

atever

-0.62

renheit

-0.60

enegger

-0.60

amber

-0.58

mercial

-0.57

ollow

-0.57

aphael

-0.56

ãĥł

-0.55

POSITIVE LOGITS

 Rapids

0.77

 Township

0.72

 County

0.70

 Heights

0.66

 Packers

0.62

 Chiefs

0.62

 Athletics

0.61

 State

0.60

 Police

0.59

apolis

0.59

Activations Density 0.265%

locations or places

geographical locations and city names

No Comments

No Known Activations