INDEX

Explanations

references to images, probably related to news or media articles

oai_token-act-pair · gpt-3.5-turbo

references to images and visual media

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

 interven

-0.76

 UNCLASSIFIED

-0.68

 intervening

-0.66

 seiz

-0.62

 abst

-0.61

 respons

-0.60

 eleph

-0.58

 complicity

-0.58

 emot

-0.58

 attent

-0.58

POSITIVE LOGITS

Steel

0.94

NFL

0.93

Coach

0.86

Atlanta

0.85

Detroit

0.83

 Cavaliers

0.83

ESPN

0.83

Pack

0.81

Denver

0.81

Philadelphia

0.79

Activations Density 0.101%

references to images, probably related to news or media articles

references to images and visual media

No Comments

No Known Activations