INDEX

Explanations

mentions of downtown areas

oai_token-act-pair · gpt-3.5-turbo

references to downtown locations

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

nat

-0.71

>>

-0.68

Analysis

-0.65

SS

-0.64

:-

-0.61

edi

-0.60

urer

-0.60

Shape

-0.59

anges

-0.59

Parent

-0.59

POSITIVE LOGITS

 downtown

3.57

 Downtown

2.21

owntown

2.11

 waterfront

1.46

 suburban

1.35

 metro

1.29

 northeast

1.24

 southeast

1.24

 courthouse

1.20

 nearby

1.19

Activations Density 0.013%

mentions of downtown areas

references to downtown locations

No Comments

No Known Activations