INDEX

Explanations

negating phrases that indicate disagreement or disapproval

oai_token-act-pair · gpt-3.5-turbo

negative statements or denials

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

éĥ

-0.80

 Inventory

-0.70

 Companies

-0.69

ership

-0.68

 Presence

-0.67

SI

-0.67

AFP

-0.66

riber

-0.66

iers

-0.66

å¥

-0.65

POSITIVE LOGITS

 necessarily

1.35

icably

1.27

icable

1.18

epad

1.15

 exactly

1.07

orious

0.98

necess

0.94

withstanding

0.94

 uncommon

0.94

 entirely

0.94

Activations Density 0.233%

negating phrases that indicate disagreement or disapproval

negative statements or denials

No Comments

No Known Activations