INDEX

Explanations

mentions of the name "j" followed by a number, which could potentially refer to a specific individual or designation

oai_token-act-pair · gpt-3.5-turbo

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 MacArthur

-0.69

houses

-0.65

 bleach

-0.64

sab

-0.61

 Gret

-0.59

 bombard

-0.59

 heck

-0.58

 dictators

-0.57

 Lauder

-0.56

pen

-0.55

POSITIVE LOGITS

unction

1.41

ournals

1.31

itsu

1.19

oint

1.18

ean

1.14

ealous

1.14

ourney

1.14

utsu

1.13

ihad

1.11

ourn

1.10

Activations Density 0.031%

mentions of the name "j" followed by a number, which could potentially refer to a specific individual or designation

No Comments

No Known Activations