INDEX

Explanations

names or words ending with "ers" and related to various contexts like people, places, and organizations

oai_token-act-pair · gpt-3.5-turbo

references to groups of people or collective identities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ADRA

-0.70

SIGN

-0.68

URRENT

-0.67

SPEC

-0.65

ATIONAL

-0.64

 forfeiture

-0.63

 numerical

-0.59

ETA

-0.58

PUT

-0.58

GoldMagikarp

-0.57

POSITIVE LOGITS

mith

1.21

haw

1.13

pace

1.10

ucker

1.10

linger

1.09

paces

1.04

cream

1.04

erker

1.03

peed

1.02

leeve

1.01

Activations Density 0.048%

names or words ending with "ers" and related to various contexts like people, places, and organizations

references to groups of people or collective identities

No Comments

No Known Activations