INDEX

Explanations

Twitter usernames/accounts

oai_token-act-pair · gpt-3.5-turbo

handles or usernames mentioned on Twitter

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

 ACTIONS

-0.72

 fingerprints

-0.68

 Islamists

-0.67

 striking

-0.65

itably

-0.64

 LIMITED

-0.63

 sights

-0.62

 diapers

-0.62

 pickups

-0.62

 CONTROL

-0.61

POSITIVE LOGITS

Bow

0.88

rick

0.87

rentice

0.85

Blog

0.83

BB

0.83

Sports

0.81

Movie

0.81

veyard

0.80

yon

0.80

brew

0.79

Activations Density 0.106%

Twitter usernames/accounts

handles or usernames mentioned on Twitter

No Comments

No Known Activations