INDEX

Explanations

phrases related to confidentiality and anonymity in text

oai_token-act-pair · gpt-3.5-turbo

references to anonymity or the desire to remain unnamed

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

 Puzzles

-0.71

nib

-0.64

Ļ½

-0.63

 lining

-0.62

HUD

-0.61

joice

-0.60

etheless

-0.60

pmwiki

-0.59

issance

-0.58

ptoms

-0.58

POSITIVE LOGITS

 anonymity

1.09

 confidentiality

0.91

 surname

0.88

 repr

0.86

 pseudonym

0.85

 confidential

0.83

 attribution

0.83

ispers

0.79

 disclosed

0.79

 disclosure

0.77

Activations Density 0.192%

phrases related to confidentiality and anonymity in text

references to anonymity or the desire to remain unnamed

No Comments

No Known Activations