INDEX

Explanations

phrases involving numerical values combined with the word "plus"

oai_token-act-pair · gpt-3.5-turbo

references to statistical data or numerical values

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

 estab

-0.71

HUM

-0.62

 revenge

-0.60

 spat

-0.59

 vengeance

-0.58

lat

-0.58

 misinterpret

-0.58

 latch

-0.58

 establishment

-0.58

 opportun

-0.57

POSITIVE LOGITS

plus

4.21

minus

2.24

Plus

2.06

 plus

1.85

 PLUS

1.57

 Plus

1.51

+.

1.35

+)

1.35

 minus

1.27

+,

1.21

Activations Density 0.010%

phrases involving numerical values combined with the word "plus"

references to statistical data or numerical values

No Comments

No Known Activations