INDEX

Explanations

phrases related to breaking or challenging established norms, rules, or structures

oai_token-act-pair · gpt-3.5-turbo

references to significant occurrences or actions related to governance and policy changes

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

00200000

-0.73

enhagen

-0.72

reau

-0.70

eers

-0.67

 Matters

-0.64

GREEN

-0.62

expr

-0.61

 offensively

-0.59

 respectfully

-0.58

enaries

-0.58

POSITIVE LOGITS

 silence

1.07

 taboo

0.98

 curfew

0.96

 seal

0.96

 barrier

0.94

 barriers

0.89

 mould

0.87

 habit

0.87

 rules

0.87

 mold

0.82

Activations Density 0.110%

phrases related to breaking or challenging established norms, rules, or structures

references to significant occurrences or actions related to governance and policy changes

No Comments

No Known Activations