INDEX

Explanations

terms related to a specific political figure, in this case, "Romney."

oai_token-act-pair · gpt-3.5-turbo

mentions of political figures, specifically Romney and Santorum

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

ĨĴ

-1.02

ivated

-0.88

 antioxid

-0.78

ĵĺ

-0.77

ives

-0.77

ivation

-0.76

omaly

-0.75

ļ

-0.73

terday

-0.73

unction

-0.73

POSITIVE LOGITS

 Romney

1.09

Romney

0.97

thal

0.91

 Voters

0.79

 Victory

0.76

Care

0.76

stein

0.73

care

0.73

 Campaign

0.72

 Biden

0.71

Activations Density 0.009%

terms related to a specific political figure, in this case, "Romney."

mentions of political figures, specifically Romney and Santorum

No Comments

No Known Activations