INDEX

Explanations

phrases related to political criticism and negative evaluation

oai_token-act-pair · gpt-3.5-turbo

references to government decisions and their consequences

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

 Folder

-0.70

 Offline

-0.70

 Finder

-0.69

 Blend

-0.68

 FANTASY

-0.66

 Cube

-0.66

 Fold

-0.64

 Explorer

-0.62

 Rookie

-0.61

 Scrolls

-0.61

POSITIVE LOGITS

 majorities

0.87

 defund

0.87

 Democr

0.84

 perpetrated

0.81

 discredited

0.79

 neocons

0.78

 threaten

0.75

milo

0.75

 pretext

0.74

 waged

0.73

Activations Density 1.495%

phrases related to political criticism and negative evaluation

references to government decisions and their consequences

No Comments

No Known Activations