INDEX

Explanations

words related to rewards, numbers, and health conditions

oai_token-act-pair · gpt-3.5-turbo

numerical data or statistics related to losses or performance metrics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ĸļ

-0.96

eanor

-0.74

isman

-0.70

older

-0.69

forcer

-0.69

iece

-0.65

emark

-0.65

 metic

-0.64

icum

-0.64

itamin

-0.64

POSITIVE LOGITS

 ones

1.27

 races

1.10

 voices

1.06

 lands

1.06

 entities

1.03

ivities

1.03

falls

1.02

 votes

1.02

 trails

1.02

 outputs

1.01

Activations Density 0.346%

words related to rewards, numbers, and health conditions

numerical data or statistics related to losses or performance metrics

No Comments

No Known Activations