INDEX

Explanations

names of political figures

oai_token-act-pair · gpt-3.5-turbo

proper nouns and names, particularly related to individuals and entities

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

 Invention

-0.67

olulu

-0.66

cffffcc

-0.65

retty

-0.62

ŃĶ

-0.59

ļéĨĴ

-0.59

};

-0.59

 Gloria

-0.56

.ï¿½

-0.56

 Palest

-0.55

POSITIVE LOGITS

 will

0.97

 intends

0.93

 could

0.93

 somehow

0.92

 might

0.89

 should

0.89

 someday

0.88

 would

0.88

 qualifies

0.86

 succeeds

0.85

Activations Density 0.479%

names of political figures

proper nouns and names, particularly related to individuals and entities

No Comments

No Known Activations