INDEX

Explanations

words related to security breaches and vulnerabilities

oai_token-act-pair · gpt-3.5-turbo

occurrences and mentions of the concept of "seconds" in various contexts

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

 mayors

-0.66

 McCartney

-0.63

 cutting

-0.59

 Leaf

-0.59

 boom

-0.58

 cane

-0.57

 bidding

-0.57

 Monster

-0.57

 Planes

-0.56

 prenatal

-0.56

POSITIVE LOGITS

sec

4.64

Sec

2.22

sec

2.04

Sec

1.68

SEC

1.59

seconds

1.35

secure

1.21

security

1.19

isec

1.18

secution

1.09

Activations Density 0.006%

words related to security breaches and vulnerabilities

occurrences and mentions of the concept of "seconds" in various contexts

No Comments

No Known Activations