INDEX

Explanations

words related to espionage and secret operations

oai_token-act-pair · gpt-3.5-turbo

terms associated with infiltration and covert operations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ruck

-0.73

--------------------

-0.73

grand

-0.68

birth

-0.68

cknowled

-0.68

-----------

-0.67

----------------

-0.67

Policy

-0.66

minus

-0.66

given

-0.66

POSITIVE LOGITS

 infiltr

1.37

 infiltration

1.29

 infiltrate

1.27

 infiltrated

1.14

iltration

0.94

 Luffy

0.75

atform

0.74

 Detection

0.73

 patrols

0.73

utsu

0.73

Activations Density 0.009%

words related to espionage and secret operations

terms associated with infiltration and covert operations

No Comments

No Known Activations