INDEX

Explanations

phrases indicating permission or action authorization

oai_token-act-pair · gpt-3.5-turbo

sentences related to permissions and what actions are allowed

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

nox

-0.83

requires

-0.74

 =================================

-0.73

quartered

-0.72

bush

-0.70

quickShipAvailable

-0.67

grain

-0.66

lda

-0.66

bol

-0.66

erg

-0.66

POSITIVE LOGITS

 practise

0.91

 carry

0.90

 waive

0.87

 waivers

0.85

 wear

0.81

 participate

0.79

 join

0.78

 Carry

0.77

 negotiate

0.76

 pass

0.75

Activations Density 0.107%

phrases indicating permission or action authorization

sentences related to permissions and what actions are allowed

No Comments

No Known Activations