INDEX

Explanations

positions or roles in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to various sports positions

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

Interstitial

-0.92

ebook

-0.86

rosso

-0.79

vous

-0.73

displayText

-0.72

cause

-0.72

actionDate

-0.70

STON

-0.69

algia

-0.68

 Flavoring

-0.67

POSITIVE LOGITS

 positions

1.11

holders

1.09

 position

1.07

holder

0.95

 holders

0.95

 holder

0.91

 vacated

0.90

ession

0.81

quo

0.78

 occupied

0.73

Activations Density 0.040%

positions or roles in various contexts

references to various sports positions

No Comments

No Known Activations