INDEX

Explanations

references to hidden or secretive information

oai_token-act-pair · gpt-3.5-turbo

themes related to secrecy and hidden information

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 TAMADRA

-0.91

 Replacement

-0.82

 upgrading

-0.79

celer

-0.77

 Upgrade

-0.76

 Combo

-0.73

 Composite

-0.73

STEM

-0.73

 Wheels

-0.72

 Playoff

-0.72

POSITIVE LOGITS

 secrecy

1.89

 anonymity

1.67

 confidentiality

1.66

 secrets

1.55

 secretive

1.46

 conceal

1.45

 confidential

1.45

 secret

1.44

 privacy

1.44

 divul

1.41

Activations Density 0.785%

references to hidden or secretive information

themes related to secrecy and hidden information

No Comments

No Known Activations