INDEX

Explanations

personal pronouns followed by verbs indicating inquiry or instructions

oai_token-act-pair · gpt-3.5-turbo

phrases focused on inquiries and dialogue about personal experiences or opinions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

BuyableInstoreAndOnline

-0.81

hement

-0.72

 largeDownload

-0.70

 Wikimedia

-0.68

vable

-0.65

swick

-0.65

ãĥ¡

-0.63

aware

-0.62

ITED

-0.62

aiman

-0.61

POSITIVE LOGITS

 bluff

0.75

Cly

0.69

 goodbye

0.67

auga

0.64

"{

0.63

ilts

0.60

Tes

0.60

DERR

0.59

gres

0.58

armac

0.58

Activations Density 0.310%

personal pronouns followed by verbs indicating inquiry or instructions

phrases focused on inquiries and dialogue about personal experiences or opinions

No Comments

No Known Activations