INDEX

Explanations

phrases related to the primary or main focus of a topic

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate the main purpose or focus of an activity or entity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

oos

-0.68

tz

-0.64

Shape

-0.61

tum

-0.60

hess

-0.59

ersen

-0.59

ety

-0.58

Tes

-0.57

aw

-0.56

 amaz

-0.56

POSITIVE LOGITS

 primarily

3.17

 mainly

2.65

 principally

2.64

 predominantly

2.34

 chiefly

2.29

 mostly

2.18

 largely

1.91

 solely

1.85

 exclusively

1.83

 predomin

1.82

Activations Density 0.013%

phrases related to the primary or main focus of a topic

phrases that indicate the main purpose or focus of an activity or entity

No Comments

No Known Activations