INDEX

Explanations

mentions of personal preferences, choices, or inclinations

oai_token-act-pair · gpt-3.5-turbo

references to individual preferences or choices

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

shut

-0.74

CHA

-0.73

lehem

-0.72

sum

-0.68

 Alam

-0.68

 shattered

-0.66

 Constable

-0.66

torn

-0.65

rave

-0.64

acre

-0.63

POSITIVE LOGITS

 preference

3.95

 preferences

3.37

 Preferences

2.36

 preferred

1.85

 preferential

1.62

 aversion

1.51

 preferring

1.44

 Pref

1.40

 prefer

1.39

 liking

1.36

Activations Density 0.017%

mentions of personal preferences, choices, or inclinations

references to individual preferences or choices

No Comments

No Known Activations