INDEX

Explanations

names of organizations or companies

oai_token-act-pair · gpt-3.5-turbo

names of organizations, locations, and titles related to official positions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 tricks

-0.82

 tack

-0.74

 brute

-0.74

zers

-0.73

 skim

-0.72

 tempting

-0.71

 curtains

-0.71

 rigging

-0.70

 headlights

-0.70

 breaches

-0.69

POSITIVE LOGITS

 Association

1.44

 Chamber

1.35

 Coalition

1.24

 Society

1.23

 Consortium

1.19

 Partnership

1.18

 Historical

1.18

 Advisory

1.17

 Advoc

1.16

 Federation

1.16

Activations Density 0.286%

names of organizations or companies

names of organizations, locations, and titles related to official positions

No Comments

No Known Activations