INDEX

Explanations

names or words related to individuals

oai_token-act-pair · gpt-3.5-turbo

instances of the word "id" and its variations, as well as the pronoun "it"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

»Ĵ

-0.74

 Skydragon

-0.72

 Haram

-0.69

 Bliss

-0.67

 Pradesh

-0.67

Pwr

-0.66

 visitation

-0.65

 Sabha

-0.62

 buffalo

-0.61

 unaff

-0.60

POSITIVE LOGITS

enthal

1.37

lein

1.31

velt

1.30

lich

1.23

enhagen

1.19

mann

1.19

linger

1.16

ler

1.14

heit

1.14

kamp

1.13

Activations Density 0.108%

names or words related to individuals

instances of the word "id" and its variations, as well as the pronoun "it"

No Comments

No Known Activations