INDEX

Explanations

the names of authors or individuals associated with written content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.10.hook_resid_post

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

conservancy

-0.75

utical

-0.73

ibly

-0.72

interstitial

-0.69

flows

-0.69

é¾įå¥ĳå£«

-0.69

conom

-0.69

cause

-0.69

Cause

-0.68

capacity

-0.67

POSITIVE LOGITS

 Anonymous

1.21

 admin

0.93

 Admin

0.91

 Guest

0.83

 Staff

0.81

hen

0.78

 Owner

0.78

 Home

0.74

jan

0.72

 john

0.71

Activations Density 0.211%

the names of authors or individuals associated with written content

No Comments

No Known Activations