INDEX

Explanations

links to shareable content or articles

oai_token-act-pair · gpt-3.5-turbo

high-activation points within disjointed structured documents

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

milo

-0.81

naire

-0.79

imens

-0.78

riott

-0.77

essee

-0.75

enegger

-0.75

 congr

-0.75

ieu

-0.72

phies

-0.72

-0.71

POSITIVE LOGITS

¿

0.79

º

0.72

¾

0.71

 Pastebin

0.71

¨

0.71

Ĺ

0.70

¦

0.70

ł

0.68

cloneembedreportprint

0.64

°

0.64

Activations Density 0.130%

links to shareable content or articles

high-activation points within disjointed structured documents

No Comments

No Known Activations