INDEX

Explanations

words related to negativity, evil, and criminal activities

oai_token-act-pair · gpt-3.5-turbo

terms related to negative or malicious attributes and actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ainers

-0.87

gain

-0.75

ersion

-0.70

 Surv

-0.70

onga

-0.69

hene

-0.68

grain

-0.68

Recomm

-0.67

Ã±a

-0.66

aple

-0.66

POSITIVE LOGITS

 plotting

1.00

 mastermind

0.95

 deed

0.94

 nefarious

0.92

 mischief

0.92

 deeds

0.92

 intent

0.86

 undermin

0.85

 sche

0.84

 havoc

0.84

Activations Density 0.054%

words related to negativity, evil, and criminal activities

terms related to negative or malicious attributes and actions

No Comments

No Known Activations