INDEX

Explanations

words related to location or organization names

oai_token-act-pair · gpt-3.5-turbo

variations of the word "well."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ItemTracker

-0.70

DragonMagazine

-0.67

olicy

-0.61

 Shinra

-0.60

 Fever

-0.60

 depreciation

-0.58

 Bahamas

-0.58

é¾įå¥ĳå£«

-0.58

 Rasm

-0.58

geist

-0.57

POSITIVE LOGITS

comed

1.31

tering

1.14

wyn

1.05

ter

1.04

estial

1.00

vin

0.97

ving

0.96

oad

0.96

tered

0.94

ves

0.94

Activations Density 0.028%

words related to location or organization names

variations of the word "well."

No Comments

No Known Activations