INDEX

Explanations

legal and policy-related terms and phrases

oai_token-act-pair · gpt-3.5-turbo

phrases related to compelling interests or justifications in policy contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 Drops

-0.76

 furiously

-0.69

 Reloaded

-0.64

 Shots

-0.63

mire

-0.63

 Slowly

-0.63

 mocked

-0.62

soon

-0.62

 Wand

-0.61

 Tackle

-0.61

POSITIVE LOGITS

reasonable

1.15

 fairness

1.09

 decency

1.06

 confidentiality

1.05

 impartial

1.05

 statutory

1.04

 ethical

1.03

 lawful

1.02

 constitutional

1.02

 equitable

1.01

Activations Density 0.611%

legal and policy-related terms and phrases

phrases related to compelling interests or justifications in policy contexts

No Comments

No Known Activations