INDEX

Explanations

references to positions or rankings on lists or charts

oai_token-act-pair · gpt-3.5-turbo

references to rankings or lists

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

nces

-0.87

nce

-0.68

 noises

-0.67

uesday

-0.65

maid

-0.62

gery

-0.61

aeus

-0.59

matter

-0.59

girl

-0.59

inous

-0.58

POSITIVE LOGITS

 list

1.34

 lists

1.17

 hierarchy

1.05

 charts

1.04

 checklist

1.03

 radar

1.03

 Lists

1.01

LIST

0.99

 blacklist

0.99

 rankings

0.98

Activations Density 0.477%

references to positions or rankings on lists or charts

references to rankings or lists

No Comments

No Known Activations