INDEX

Explanations

references to news articles and their publication details such as dates, authors, and news outlets

oai_token-act-pair · gpt-3.5-turbo

references to image or photo credits in a document

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

 pronoun

-0.79

 favour

-0.79

 repe

-0.73

 demol

-0.73

 boun

-0.73

 follows

-0.73

 bonding

-0.71

 shaming

-0.71

 tender

-0.71

 shame

-0.71

POSITIVE LOGITS

Getty

1.51

AFP

1.47

Flickr

1.46

Associated

1.44

Reuters

1.41

Bloomberg

1.35

Courtesy

1.31

EPA

1.27

CBC

1.25

USA

1.17

Activations Density 0.030%

references to news articles and their publication details such as dates, authors, and news outlets

references to image or photo credits in a document

No Comments

No Known Activations