INDEX

Explanations

terms related to social justice and political discussions, particularly focusing on race and historical figures

oai_token-act-pair · gpt-3.5-turbo

mentions of political ideologies and discussions around social identity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 util

-0.89

-0.82

 didnt

-0.81

Â´

-0.77

lol

-0.74

 Clan

-0.73

 whilst

-0.73

 confir

-0.73

!!

-0.73

 stating

-0.72

POSITIVE LOGITS

NPR

1.26

 Slate

1.22

 POLITICO

1.20

 Brookings

1.14

 HuffPost

1.12

 Pulitzer

1.02

 Economist

1.02

NPR

0.99

 Smithsonian

0.99

Vox

0.97

Activations Density 0.459%

terms related to social justice and political discussions, particularly focusing on race and historical figures

mentions of political ideologies and discussions around social identity

No Comments

No Known Activations