INDEX

Explanations

references to availability or access

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the availability of resources or information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

²¾

-0.70

mot

-0.64

asus

-0.63

iosity

-0.62

isol

-0.62

telling

-0.60

mouse

-0.59

mother

-0.59

kos

-0.58

ICO

-0.58

POSITIVE LOGITS

 online

0.86

 unrestricted

0.76

via

0.74

 anywhere

0.73

for

0.73

 through

0.72

 digitally

0.71

oin

0.70

 locally

0.70

 cheaply

0.70

Activations Density 0.069%

references to availability or access

phrases indicating the availability of resources or information

No Comments

No Known Activations