INDEX

Explanations

phrases related to specific geographic locations, particularly places with the word "anga" in the name

oai_token-act-pair · gpt-3.5-turbo

references to manga or related content in the text

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Patriots

-0.65

 exact

-0.64

 prob

-0.63

 toler

-0.63

 ident

-0.63

-0.60

 calls

-0.59

 call

-0.59

ser

-0.59

 doub

-0.59

POSITIVE LOGITS

anga

4.66

anges

1.37

 Manga

1.29

angan

1.28

akura

1.23

anka

1.16

oola

1.12

ahime

1.12

ang

1.10

 manga

1.07

Activations Density 0.013%

phrases related to specific geographic locations, particularly places with the word "anga" in the name

references to manga or related content in the text

No Comments

No Known Activations