INDEX

Explanations

phrases related to investigations, including investigator titles and actions

oai_token-act-pair · gpt-3.5-turbo

references to various types of investigators and investigative roles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

dos

-1.00

usable

-0.76

Tea

-0.70

election

-0.68

amping

-0.68

zl

-0.68

perm

-0.68

hur

-0.68

days

-0.67

carry

-0.66

POSITIVE LOGITS

 investigator

1.27

 Investigator

1.03

 investigates

0.79

 examiner

0.77

Reviewer

0.76

 explorer

0.76

 investig

0.76

 Investigation

0.74

 reviewer

0.74

 Mechan

0.73

Activations Density 0.008%

phrases related to investigations, including investigator titles and actions

references to various types of investigators and investigative roles

No Comments

No Known Activations