INDEX

Explanations

ideologically-charged terms related to political ideology and activism

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 increments

-0.77

Delivery

-0.76

angan

-0.76

 bilateral

-0.72

Owner

-0.70

ãĥ¼ãĥ³

-0.68

Ī

-0.65

shown

-0.63

 Bulldogs

-0.62

eway

-0.62

POSITIVE LOGITS

ervatives

1.44

ervative

1.35

 alike

1.09

 rejoice

1.08

paces

1.07

 everywhere

1.05

who

1.04

peak

1.00

hip

0.92

 Anonymous

0.89

Activations Density 0.199%

ideologically-charged terms related to political ideology and activism

No Comments

No Known Activations