INDEX

Explanations

references to individuals holding past positions or titles

oai_token-act-pair · gpt-3.5-turbo

terms related to individuals with former roles in specific positions or organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ups

-0.71

 ling

-0.64

 purposes

-0.62

 crops

-0.61

atching

-0.58

 dividends

-0.58

 bleed

-0.57

 exped

-0.57

 overd

-0.57

 knit

-0.56

POSITIVE LOGITS

former

4.59

formation

1.29

forming

1.24

vice

1.14

older

1.08

form

1.07

chief

1.04

FORM

1.03

creator

1.03

forms

1.02

Activations Density 0.020%

references to individuals holding past positions or titles

terms related to individuals with former roles in specific positions or organizations

No Comments

No Known Activations