INDEX

Explanations

terms related to government and policy

oai_token-act-pair · gpt-3.5-turbo

references to government and its activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 traumatic

-0.72

terday

-0.65

 moms

-0.65

ÄŁ

-0.63

 grandma

-0.62

 derail

-0.61

SPONSORED

-0.61

 queens

-0.59

imeters

-0.59

ĻĤ

-0.59

POSITIVE LOGITS

 Accountability

1.21

 Order

0.90

 Department

0.90

 Communications

0.89

 Manager

0.89

 Authority

0.88

 Office

0.87

 Builder

0.87

 Relations

0.87

 Council

0.87

Activations Density 0.227%

terms related to government and policy

references to government and its activities

No Comments

No Known Activations