INDEX

Explanations

specific mentions of geographical locations, specifically with the word 'Creek'

oai_token-act-pair · gpt-3.5-turbo

references to various "Creek" locations or waterways

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

berman

-0.77

 newsp

-0.70

emort

-0.63

instein

-0.62

 occupational

-0.62

 explanatory

-0.61

mble

-0.61

innie

-0.61

neapolis

-0.60

[_

-0.60

POSITIVE LOGITS

 Creek

1.18

dale

1.11

weed

0.97

 Canyon

0.93

waters

0.92

bank

0.83

 Valley

0.83

cliffe

0.82

 Falls

0.82

 River

0.82

Activations Density 0.014%

specific mentions of geographical locations, specifically with the word 'Creek'

references to various "Creek" locations or waterways

No Comments

No Known Activations