INDEX

Explanations

references to reviewing documents or information

oai_token-act-pair · gpt-3.5-turbo

references to various forms of examination and analysis

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

Ń·

-0.67

cott

-0.67

ario

-0.66

å§

-0.64

 hugs

-0.63

ajor

-0.63

utters

-0.62

sonian

-0.62

udeb

-0.62

POSE

-0.62

POSITIVE LOGITS

 closely

1.00

 carefully

0.93

 objectively

0.92

 archives

0.89

 aloud

0.86

 trends

0.83

 databases

0.80

 favorably

0.78

geist

0.77

 unfold

0.77

Activations Density 0.704%

references to reviewing documents or information

references to various forms of examination and analysis

No Comments

No Known Activations