INDEX

Explanations

words related to security and protection

oai_token-act-pair · gpt-3.5-turbo

references to various forms of protection, particularly related to legal and regulatory contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 spoon

-0.69

aleb

-0.68

score

-0.67

ansky

-0.67

 salsa

-0.66

 busy

-0.65

 beans

-0.65

 beet

-0.64

otto

-0.64

sem

-0.64

POSITIVE LOGITS

 Protection

3.94

 protection

2.33

protection

2.13

 Protect

1.77

 Prevention

1.72

 Protective

1.70

Protect

1.69

 protections

1.57

 PROT

1.53

Prot

1.40

Activations Density 0.016%

words related to security and protection

references to various forms of protection, particularly related to legal and regulatory contexts

No Comments

No Known Activations