INDEX

Explanations

descriptions of people's behaviors or opinions

oai_token-act-pair · gpt-3.5-turbo

references to people's opinions and reactions

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

 srfAttach

-1.07

-+

-0.71

Delivery

-0.69

 Alone

-0.69

oken

-0.63

 Tours

-0.62

Vs

-0.61

 equals

-0.61

 Settlement

-0.61

ãĥĺ

-0.61

POSITIVE LOGITS

 argue

1.43

 speculate

1.37

 speculated

1.30

 theor

1.27

 accuse

1.25

 errone

1.24

 criticize

1.23

 wondered

1.23

 mistakenly

1.22

 misinterpret

1.19

Activations Density 0.233%

descriptions of people's behaviors or opinions

references to people's opinions and reactions

No Comments

No Known Activations