INDEX

Explanations

references to specific entities, such as companies, countries, teams, players, or games

oai_token-act-pair · gpt-3.5-turbo

mentions of entities such as companies, countries, teams, and games

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

ories

-0.67

 puppies

-0.64

 trophies

-0.64

elight

-0.63

 Mans

-0.62

 passports

-0.62

 bolts

-0.62

latest

-0.61

Kid

-0.61

etsk

-0.61

POSITIVE LOGITS

 Called

0.94

ogram

0.78

alyst

0.76

onym

0.73

atical

0.72

ribe

0.71

 whose

0.70

 Named

0.69

 scientist

0.69

ulent

0.69

Activations Density 0.361%

references to specific entities, such as companies, countries, teams, players, or games

mentions of entities such as companies, countries, teams, and games

No Comments

No Known Activations