INDEX

Explanations

mentions of percentages

oai_token-act-pair · gpt-3.5-turbo

references to demographics or population statistics

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

 sunset

-0.62

 Ares

-0.55

 Chimera

-0.55

ciation

-0.54

 Prometheus

-0.53

 Saturn

-0.52

 Archdemon

-0.50

 Scorpion

-0.50

 Tornado

-0.50

 Inferno

-0.49

POSITIVE LOGITS

who

0.74

selves

0.73

 backgrounds

0.70

 whom

0.69

doms

0.69

pired

0.67

heastern

0.66

 sworn

0.64

 surn

0.63

cler

0.61

Activations Density 1.898%

mentions of percentages

references to demographics or population statistics

No Comments

No Known Activations