INDEX

Explanations

names of anime series

oai_token-act-pair · gpt-3.5-turbo

references to a specific anime series and its associated content

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Pool

-0.68

 Liver

-0.67

 sourcing

-0.65

fighters

-0.63

 Benz

-0.63

zyk

-0.63

 Caribbean

-0.62

 Lamar

-0.60

 Tomas

-0.60

 Finn

-0.59

POSITIVE LOGITS

ga

3.86

 adjourn

1.43

 lect

1.31

Contents

1.29

ni

1.20

 sequest

1.05

grave

1.04

 ceremonial

1.03

 filib

1.01

ka

1.00

Activations Density 0.036%

names of anime series

references to a specific anime series and its associated content

No Comments

No Known Activations