INDEX

Explanations

terms related to security personnel or law enforcement officers

oai_token-act-pair · gpt-3.5-turbo

references to security personnel

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

BLE

-0.74

 Scotia

-0.70

ãĥ£

-0.67

FORM

-0.66

NOT

-0.64

astical

-0.63

KEN

-0.63

ctive

-0.63

 Hemp

-0.61

à¤

-0.60

POSITIVE LOGITS

men

1.07

 guarding

1.04

andowski

0.97

 guards

0.94

maid

0.92

guards

0.90

 stationed

0.88

hip

0.85

rail

0.83

masters

0.81

Activations Density 0.009%

terms related to security personnel or law enforcement officers

references to security personnel

No Comments

No Known Activations