INDEX

Explanations

instances where the reader is asked to take action or make a decision

oai_token-act-pair · gpt-3.5-turbo

direct references to the audience or reader

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

iform

-0.74

 wait

-0.68

Cooldown

-0.62

inian

-0.62

ominated

-0.59

 rainy

-0.58

 Rampage

-0.58

 Needs

-0.57

 Usage

-0.57

POSITIVE LOGITS

 accompl

0.84

 unwittingly

0.83

 depri

0.83

resents

0.81

 dece

0.77

 gain

0.77

 implicitly

0.76

 embold

0.75

 circumvent

0.74

 meant

0.74

Activations Density 0.262%

instances where the reader is asked to take action or make a decision

direct references to the audience or reader

No Comments

No Known Activations