INDEX

Explanations

phrases related to directing or sending information

oai_token-act-pair · gpt-3.5-turbo

references to direct communication or messages

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

 cautiously

-0.75

etheless

-0.70

manship

-0.65

uly

-0.65

assi

-0.63

ply

-0.62

 brightest

-0.61

ĻĤ

-0.61

pse

-0.61

 strategically

-0.61

POSITIVE LOGITS

SpaceEngineers

0.73

achable

0.71

govtrack

0.68

hyde

0.68

UGC

0.64

ulkan

0.63

ADA

0.63

ierrez

0.62

enegger

0.62

arez

0.62

Activations Density 0.146%

phrases related to directing or sending information

references to direct communication or messages

No Comments

No Known Activations