INDEX

Explanations

instances of permission being sought or granted

oai_token-act-pair · gpt-3.5-turbo

references to permission or requests for authorization

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

 Scream

-0.77

rued

-0.75

ulz

-0.71

ulative

-0.68

DeV

-0.67

urgy

-0.66

tal

-0.66

oche

-0.66

EMS

-0.66

iked

-0.65

POSITIVE LOGITS

 permission

1.50

 permissions

1.21

 waivers

0.90

 approvals

0.89

 granted

0.85

 confir

0.84

 clearance

0.84

 license

0.83

 licence

0.81

 authorization

0.81

Activations Density 0.007%

instances of permission being sought or granted

references to permission or requests for authorization

No Comments

No Known Activations