INDEX

Explanations

personal pronouns followed by forms of the verb "ask."

oai_token-act-pair · gpt-3.5-turbo

pronouns and references to self or others in queries or requests

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

inction

-0.73

ushima

-0.72

ngth

-0.70

olate

-0.69

acca

-0.66

piracy

-0.66

olated

-0.65

 Tracks

-0.64

oneliness

-0.62

BuyableInstoreAndOnline

-0.62

POSITIVE LOGITS

 forgiveness

0.86

DERR

0.86

 probing

0.85

 rhet

0.84

 questions

0.83

 permission

0.75

 daddy

0.74

 farewell

0.72

 question

0.70

 plaint

0.70

Activations Density 0.157%

personal pronouns followed by forms of the verb "ask."

pronouns and references to self or others in queries or requests

No Comments

No Known Activations