INDEX

Explanations

security-related terms and concepts, such as defense, attack, firewall, and mitigation

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

Data Type

torch.float32

Hook Point

blocks.8.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

lore

-0.68

zos

-0.66

hall

-0.63

mitt

-0.62

Bee

-0.61

liam

-0.60

 Parenthood

-0.58

wheel

-0.57

 Kinn

-0.57

cow

-0.56

POSITIVE LOGITS

 against

1.13

 Against

0.93

against

0.93

iveness

0.92

ously

0.91

ively

0.88

 perimeter

0.79

atively

0.78

folios

0.78

heed

0.77

Activations Density 2.130%

security-related terms and concepts, such as defense, attack, firewall, and mitigation

No Comments

No Known Activations