INDEX

Explanations

proper nouns related to media or news outlets

oai_token-act-pair · gpt-3.5-turbo

references to content or features related to newsletters

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

Sov

-0.69

lly

-0.68

kefeller

-0.65

cest

-0.64

ufact

-0.64

nai

-0.63

kh

-0.63

pac

-0.62

#$

-0.62

heit

-0.62

POSITIVE LOGITS

âĶĢâĶĢâĶĢâĶĢ

0.63

 sidebar

0.60

interstitial

0.57

 âĸ

0.56

 iTunes

0.56

 ARTICLE

0.54

qus

0.54

 CrossRef

0.54

 âĨ

0.54

 reddit

0.54

Activations Density 0.042%

proper nouns related to media or news outlets

references to content or features related to newsletters

No Comments

No Known Activations