INDEX

Explanations

user identifiers or hashtags

oai_token-act-pair · gpt-3.5-turbo

hashtags or labels used in social media posts

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

 pestic

-0.72

Sov

-0.70

ahime

-0.70

veh

-0.70

 weap

-0.69

stitial

-0.69

 horizont

-0.68

 Wander

-0.67

 colle

-0.66

 predec

-0.66

POSITIVE LOGITS

########

1.30

################################

1.29

################

1.14

###

0.91

##

0.82

 Posts

0.82

nice

0.79

region

0.78

bitcoin

0.71

deck

0.71

Activations Density 0.007%

user identifiers or hashtags

hashtags or labels used in social media posts

No Comments

No Known Activations