INDEX

Explanations

websites where articles were originally published

oai_token-act-pair · gpt-3.5-turbo

references to article origins and attributions

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

cffffcc

-0.72

anything

-0.69

oute

-0.67

auc

-0.66

lob

-0.65

beit

-0.64

ients

-0.64

 indiscrim

-0.62

oles

-0.61

nce

-0.60

POSITIVE LOGITS

 POLITICO

1.14

 HuffPost

1.10

 VICE

1.09

 TIME

1.01

 Investor

0.91

 Forbes

0.90

 Alter

0.88

 Huffington

0.87

 Tablet

0.87

 Bloomberg

0.85

Activations Density 0.120%

websites where articles were originally published

references to article origins and attributions

No Comments

No Known Activations