INDEX

Explanations

concepts or ideas that are defining or have the potential to define a particular topic or subject

oai_token-act-pair · gpt-3.5-turbo

instances of the word "define" or related concepts of definition

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Saints

-0.79

itch

-0.73

ience

-0.70

rentice

-0.68

 Patron

-0.63

Sor

-0.62

bay

-0.62

 Meadows

-0.61

 Plants

-0.61

 Parish

-0.61

POSITIVE LOGITS

 defines

1.17

initions

1.16

 define

1.07

inition

1.02

 defined

0.92

 defining

0.91

generic

0.89

define

0.89

 definitions

0.88

 Definitions

0.81

Activations Density 0.010%

concepts or ideas that are defining or have the potential to define a particular topic or subject

instances of the word "define" or related concepts of definition

No Comments

No Known Activations