INDEX

Explanations

references to political or military discussions

oai_token-act-pair · gpt-3.5-turbo

ellipses or pauses in the text, indicating omissions or continuations

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

 Marketable

-0.87

uploads

-0.84

culosis

-0.75

phrine

-0.70

å¤©

-0.67

å¼

-0.67

-0.66

èĢħ

-0.65

 Shattered

-0.64

Discussion

-0.64

POSITIVE LOGITS

..

0.95

Ħ¢

0.91

imately

0.70

eme

0.69

 fossils

0.65

rosc

0.65

iami

0.65

eneg

0.65

alde

0.64

dur

0.62

Activations Density 0.011%

references to political or military discussions

ellipses or pauses in the text, indicating omissions or continuations

No Comments

No Known Activations