INDEX

Explanations

terms related to the person Edward Snowden

oai_token-act-pair · gpt-3.5-turbo

references to Edward Snowden

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

sembly

-0.78

rontal

-0.73

ogun

-0.73

reek

-0.73

ournals

-0.71

hetical

-0.70

opathic

-0.70

ombo

-0.69

hesis

-0.69

ptive

-0.69

POSITIVE LOGITS

 Snowden

1.01

DragonMagazine

0.81

 trove

0.76

 Greenwald

0.76

 Files

0.75

 whistleblowers

0.75

 Archive

0.73

 revelations

0.73

 whistleblower

0.71

bin

0.68

Activations Density 0.014%

terms related to the person Edward Snowden

references to Edward Snowden

No Comments

No Known Activations