INDEX

Explanations

proper nouns or names, particularly "Kaine"

oai_token-act-pair · gpt-3.5-turbo

references to the character Kaine in a narrative context

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

 Lumpur

-0.96

nesium

-0.79

oros

-0.76

igm

-0.71

plex

-0.71

 Institution

-0.70

keye

-0.70

immer

-0.70

hook

-0.69

ickson

-0.68

POSITIVE LOGITS

 Kaine

0.91

 vans

0.70

 Mercenary

0.69

 delinqu

0.68

INTON

0.68

ibilities

0.68

sters

0.67

 prosecut

0.66

 Hebdo

0.65

Activations Density 0.057%

proper nouns or names, particularly "Kaine"

references to the character Kaine in a narrative context

No Comments

No Known Activations