INDEX

Explanations

references to media coverage and misinformation

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 exerc

-0.76

 Solitaire

-0.73

apesh

-0.70

maxwell

-0.68

vi

-0.65

 Mobility

-0.65

ultimate

-0.65

 unlocked

-0.64

coh

-0.64

 Calculator

-0.63

POSITIVE LOGITS

 portraying

1.19

reporting

1.08

 sensational

1.07

 reporting

1.02

 portray

1.01

 biased

0.98

 airing

0.98

 outlets

0.98

 Reporting

0.96

 covering

0.95

Activations Density 0.585%

references to media coverage and misinformation

No Comments

No Known Activations