INDEX

Explanations

instances of being physically confined or restricted

oai_token-act-pair · gpt-3.5-turbo

instances of the word "locked" and its variations, indicating confinement or restriction

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

 exaggeration

-0.72

ahon

-0.70

 Interpret

-0.69

aste

-0.68

Footnote

-0.67

plot

-0.66

 Insp

-0.65

 Cosponsors

-0.65

eness

-0.65

brate

-0.65

POSITIVE LOGITS

 locked

3.45

 Locked

2.22

 lock

2.06

locked

2.01

 locking

2.01

 unlocked

1.97

 locks

1.91

Lock

1.54

 chained

1.51

 Lock

1.46

Activations Density 0.018%

instances of being physically confined or restricted

instances of the word "locked" and its variations, indicating confinement or restriction

No Comments

No Known Activations