INDEX

Explanations

documents and statements sent or issued by specific individuals or offices

oai_token-act-pair · gpt-3.5-turbo

phrases indicating authorship or sources of communication, particularly letters or statements

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

iencies

-0.89

}}}

-0.78

selection

-0.71

ikes

-0.70

ipedia

-0.68

uality

-0.68

tle

-0.68

cipline

-0.67

GET

-0.67

ually

-0.65

POSITIVE LOGITS

 Cyrus

0.77

 sender

0.68

 Archbishop

0.66

 governors

0.66

CCP

0.65

 Maurice

0.65

 Governor

0.65

 Joel

0.65

 petitioner

0.64

 Director

0.64

Activations Density 0.163%

documents and statements sent or issued by specific individuals or offices

phrases indicating authorship or sources of communication, particularly letters or statements

No Comments

No Known Activations