INDEX

Explanations

government officials by their titles

oai_token-act-pair · gpt-3.5-turbo

references to political figures, specifically governors and senators

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

selves

-0.74

OUND

-0.64

 Australians

-0.63

ciation

-0.62

 Siber

-0.62

ateral

-0.62

axis

-0.60

 embodiments

-0.60

ÙĴ

-0.59

 Retrieved

-0.58

POSITIVE LOGITS

aucuses

0.76

stand

0.73

 Nicholas

0.71

itol

0.70

 caucuses

0.69

annis

0.67

abulary

0.67

eanor

0.67

Xan

0.66

ģ«

0.64

Activations Density 0.098%

government officials by their titles

references to political figures, specifically governors and senators

No Comments

No Known Activations