INDEX

Explanations

people in authoritative positions or organizations

oai_token-act-pair · gpt-3.5-turbo

phrases referencing organizational roles or titles

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

 distractions

-0.73

 knees

-0.68

NetMessage

-0.68

 basics

-0.66

 coales

-0.62

 amounts

-0.60

 wrink

-0.59

 figur

-0.59

theless

-0.59

 expans

-0.59

POSITIVE LOGITS

 whom

0.79

NCT

0.71

 Telecom

0.68

 Nottingham

0.66

inance

0.65

 Flags

0.64

MI

0.64

 Fired

0.63

efficients

0.63

RIP

0.62

Activations Density 0.106%

people in authoritative positions or organizations

phrases referencing organizational roles or titles

No Comments

No Known Activations