INDEX

Explanations

terms related to breaches, violations, frauds, or misconduct

oai_token-act-pair · gpt-3.5-turbo

instances of the word "breach" and its variations, indicating violations or infractions

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

oline

-0.72

 Color

-0.72

olin

-0.69

 educated

-0.67

 Pres

-0.64

 Analog

-0.64

comed

-0.64

 offic

-0.63

Alf

-0.63

 coloring

-0.62

POSITIVE LOGITS

 breach

4.05

 breaches

3.02

 Breach

2.56

 breached

2.54

 breaching

2.39

 rupture

1.68

 violation

1.46

 intrusion

1.38

 lapse

1.36

 leak

1.26

Activations Density 0.011%

terms related to breaches, violations, frauds, or misconduct

instances of the word "breach" and its variations, indicating violations or infractions

No Comments

No Known Activations