INDEX

Explanations

proper nouns or names, particularly related to people and places

oai_token-act-pair · gpt-3.5-turbo

names of geographical locations and their associations

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

 Astron

-0.69

 Scourge

-0.69

Used

-0.69

Know

-0.69

 Breach

-0.68

OTUS

-0.67

 Secrets

-0.66

 Nanto

-0.65

 Rapt

-0.65

Wars

-0.64

POSITIVE LOGITS

adal

2.84

amba

2.02

amide

1.66

oyle

1.51

aram

1.28

anka

1.28

ilic

1.22

ansson

1.19

olson

1.09

iesel

1.05

Activations Density 0.048%

proper nouns or names, particularly related to people and places

names of geographical locations and their associations

No Comments

No Known Activations