INDEX

Explanations

proper nouns, specifically the name "Derek"

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Derek" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 loop

-0.81

 overflow

-0.80

bus

-0.75

 vill

-0.71

NAT

-0.70

Rat

-0.70

 Gent

-0.68

 Witch

-0.68

hop

-0.67

 loops

-0.66

POSITIVE LOGITS

 Derek

3.34

erek

2.20

 Darren

1.08

 Clyde

1.07

 Raiders

1.06

 Kelvin

1.03

 Draper

1.01

 Raider

0.99

 Khal

0.99

 Manziel

0.97

Activations Density 0.030%

proper nouns, specifically the name "Derek"

mentions of the name "Derek" in various contexts

No Comments

No Known Activations