INDEX

Explanations

references to specific topics or entities, such as "Resident Evil" or "Alone"

oai_token-act-pair · gpt-3.5-turbo

terms related to solitude and residency

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

tones

-0.82

tip

-0.68

 wrinkles

-0.67

 corners

-0.66

orted

-0.65

 mockery

-0.64

 folds

-0.64

hol

-0.62

 rounded

-0.62

§

-0.62

POSITIVE LOGITS

 Alone

3.29

 Resident

1.42

 Already

1.21

 absentee

1.21

 Guilty

1.12

 Absent

0.99

 Nightmares

0.96

Already

0.90

hower

0.85

 Logged

0.83

Activations Density 0.025%

references to specific topics or entities, such as "Resident Evil" or "Alone"

terms related to solitude and residency

No Comments

No Known Activations