INDEX

Explanations

phrases related to figures or entities with the initial "j"

oai_token-act-pair · gpt-3.5-turbo

repeated references to the letter "j."

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

 Lauder

-0.74

BALL

-0.71

 dispers

-0.69

 bleach

-0.68

houses

-0.65

holders

-0.64

 productive

-0.62

 dispersed

-0.62

 rooting

-0.62

 outgoing

-0.62

POSITIVE LOGITS

ournals

1.42

acket

1.28

ealous

1.28

itsu

1.23

unction

1.19

utsu

1.14

ordan

1.13

ihad

1.11

unal

1.10

oint

1.08

Activations Density 0.031%

phrases related to figures or entities with the initial "j"

repeated references to the letter "j."

No Comments

No Known Activations