INDEX

Explanations

terms related to malicious software, especially malware

oai_token-act-pair · gpt-3.5-turbo

references to malware and ransomware

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

-0.82

ffen

-0.79

ilts

-0.76

eree

-0.71

rite

-0.70

arist

-0.70

 Duty

-0.68

infeld

-0.66

lot

-0.65

 Attend

-0.65

POSITIVE LOGITS

bytes

1.05

 malware

1.02

eware

0.93

ware

0.87

 infection

0.86

 infections

0.85

 infect

0.84

 Mehran

0.80

 outbreak

0.77

 ransomware

0.77

Activations Density 0.019%

terms related to malicious software, especially malware

references to malware and ransomware

No Comments

No Known Activations