INDEX

Explanations

mentions of violent or tense situations involving specific characters

oai_token-act-pair · gpt-3.5-turbo

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

 Abedin

-0.94

cham

-0.86

airo

-0.80

 chrom

-0.79

 Skype

-0.78

 Esper

-0.77

Tab

-0.75

 triangle

-0.72

UTC

-0.72

IE

-0.72

POSITIVE LOGITS

 Rick

2.05

Rick

2.03

 Morty

1.70

 Carl

1.21

 Hick

1.13

 Daryl

1.05

rick

1.04

 Ricky

1.02

 Vick

1.01

 Rept

0.97

Activations Density 0.187%

mentions of violent or tense situations involving specific characters

No Comments

No Known Activations