INDEX

Explanations

legal and governmental terms and phrases

oai_token-act-pair · gpt-3.5-turbo

quantities or numerical data related to events and actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

laughs

-0.47

 fucking

-0.47

 crappy

-0.46

Laughs

-0.46

 simulac

-0.44

 goddamn

-0.44

 queens

-0.43

 somew

-0.42

Obi

-0.42

 Sorce

-0.42

POSITIVE LOGITS

 FEMA

0.48

announced

0.46

reporting

0.45

 UNESCO

0.45

 insurers

0.45

 citing

0.44

 accredited

0.43

gerald

0.43

 issuance

0.42

DOJ

0.42

Activations Density 4.631%

legal and governmental terms and phrases

quantities or numerical data related to events and actions

No Comments

No Known Activations