INDEX

Explanations

phrases related to the intention or purpose of an action towards a specific target

oai_token-act-pair · gpt-3.5-turbo

phrases indicating intent or purpose

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

 IMAGES

-0.64

å£«

-0.63

anon

-0.62

 Joined

-0.61

 Situation

-0.60

 Transcript

-0.59

etc

-0.59

 Frie

-0.58

 Operating

-0.57

apt

-0.54

POSITIVE LOGITS

 squarely

1.21

 specifically

0.91

at

0.90

 towards

0.88

 toward

0.86

 aimed

0.86

 primarily

0.84

 solely

0.81

 principally

0.80

 mainly

0.78

Activations Density 0.056%

phrases related to the intention or purpose of an action towards a specific target

phrases indicating intent or purpose

No Comments

No Known Activations