INDEX

Explanations

politicians and their respective states and parties

oai_token-act-pair · gpt-3.5-turbo

negative political sentiment

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

DragonMagazine

-0.87

llah

-0.85

obyl

-0.83

urden

-0.66

tics

-0.62

 radar

-0.61

htaking

-0.61

 cache

-0.61

 icing

-0.61

 layer

-0.60

POSITIVE LOGITS

adj

1.02

nom

0.98

California

0.95

Calif

0.91

member

0.87

affiliated

0.86

ranked

0.84

prime

0.81

GG

0.81

degree

0.80

Activations Density 0.044%

politicians and their respective states and parties

negative political sentiment

No Comments

No Known Activations