INDEX

Explanations

numbers related to ranking or counting

oai_token-act-pair · gpt-3.5-turbo

references to specific numerical ratings or rankings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 crawl

-0.73

offer

-0.65

 environment

-0.62

ãĤ¹ãĥĪ

-0.62

 redemption

-0.62

 deeds

-0.60

 conditioning

-0.58

om

-0.58

 condition

-0.56

DMV

-0.56

POSITIVE LOGITS

teen

2.24

teenth

1.50

eteen

1.41

een

1.33

eenth

1.19

ths

1.04

eties

1.02

eteenth

0.99

fif

0.98

ety

0.96

Activations Density 0.017%

numbers related to ranking or counting

references to specific numerical ratings or rankings

No Comments

No Known Activations