INDEX

Explanations

phrases related to politics and government officials/situations

oai_token-act-pair · gpt-3.5-turbo

references to political figures and organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

Redditor

-0.79

 moms

-0.74

.ï¿½

-0.72

\"

-0.70

ADVERTISEMENT

-0.70

mom

-0.67

ursday

-0.67

keyes

-0.67

ï¸ı

-0.66

.","

-0.65

POSITIVE LOGITS

 recognised

1.18

 criticised

1.14

 (Â£

1.09

 hitherto

1.08

 analysed

1.04

 Nato

1.03

 favourable

1.03

 apologised

1.03

 organised

1.02

 authorised

1.00

Activations Density 0.638%

phrases related to politics and government officials/situations

references to political figures and organizations

No Comments

No Known Activations