INDEX

Explanations

references to the newest or most recent items or releases

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "newness" or "new" items and experiences

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

Scotland

-0.65

bank

-0.61

nikov

-0.61

 England

-0.60

Â£

-0.58

IRA

-0.58

England

-0.57

 Correspond

-0.57

dom

-0.57

tery

-0.57

POSITIVE LOGITS

 newest

2.85

 newer

2.76

new

1.64

 latest

1.54

 newcomers

1.47

 newcomer

1.46

new

1.42

NEW

1.35

 newly

1.27

latest

1.17

Activations Density 0.013%

references to the newest or most recent items or releases

references to the concept of "newness" or "new" items and experiences

No Comments

No Known Activations