INDEX

Explanations

names of fictional locations and ranks

oai_token-act-pair · gpt-3.5-turbo

references to characters or titles from a fantasy narrative

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Zucker

-0.94

ilater

-0.88

ierrez

-0.86

 horizont

-0.85

ï¸

-0.81

emade

-0.78

ï¸ı

-0.78

 POLIT

-0.75

mercial

-0.74

 diabetic

-0.72

POSITIVE LOGITS

sword

1.35

 Elves

1.27

Wyr

1.21

 Archdemon

1.18

 orcs

1.15

 dwarves

1.14

 Dagger

1.13

 mages

1.12

 sorce

1.11

 dragons

1.10

Activations Density 0.424%

names of fictional locations and ranks

references to characters or titles from a fantasy narrative

No Comments

No Known Activations