INDEX

Explanations

words related to doing work or tasks

oai_token-act-pair · gpt-3.5-turbo

repeated references to the concept of labor or employment

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ogle

-0.69

agonist

-0.64

ername

-0.64

ãĥĥãĥī

-0.63

 invaded

-0.62

 Ratings

-0.61

icer

-0.60

ceivable

-0.60

KER

-0.60

¹

-0.60

POSITIVE LOGITS

 work

3.46

 Work

2.46

work

2.46

 WORK

2.33

Work

2.20

 works

1.98

 worked

1.74

WORK

1.61

 working

1.59

working

1.51

Activations Density 0.054%

words related to doing work or tasks

repeated references to the concept of labor or employment

No Comments

No Known Activations