INDEX

Explanations

words related to legal or formal language, such as titles, names, and legal terms

oai_token-act-pair · gpt-3.5-turbo

abbreviations and acronyms related to organizations or titles

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

ttes

-0.72

icult

-0.70

analy

-0.70

itone

-0.65

atography

-0.65

iland

-0.64

reys

-0.63

ano

-0.62

ships

-0.62

estern

-0.62

POSITIVE LOGITS

volent

0.75

 retribution

0.66

 Caesar

0.64

 Horus

0.64

 wrath

0.64

 emperor

0.61

 CHRIST

0.61

 holiest

0.58

 jihad

0.58

 resemb

0.58

Activations Density 0.266%

words related to legal or formal language, such as titles, names, and legal terms

abbreviations and acronyms related to organizations or titles

No Comments

No Known Activations