INDEX

Explanations

names of books or titles of works

oai_token-act-pair · gpt-3.5-turbo

names of books and literary works

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 parity

-0.76

 allowances

-0.74

 impair

-0.74

 cones

-0.74

 deliveries

-0.70

aspers

-0.69

 statements

-0.68

anced

-0.67

 restrictive

-0.66

NetMessage

-0.65

POSITIVE LOGITS

 Trilogy

1.20

 Podcast

1.05

 Brigade

1.02

 Collection

1.01

 Massacre

0.99

mma

0.97

 Awakens

0.97

 Problem

0.95

 Handbook

0.94

 Chronicle

0.93

Activations Density 0.418%

names of books or titles of works

names of books and literary works

No Comments

No Known Activations