INDEX

Explanations

concepts related to security and caution

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

herty

-0.61

Chapter

-0.60

uran

-0.59

avin

-0.55

uum

-0.55

atform

-0.55

Documents

-0.55

-0.54

ool

-0.54

ampions

-0.53

POSITIVE LOGITS

to

1.27

to

1.27

TO

1.03

TO

1.00

To

0.96

To

0.96

 thereto

0.95

 unto

0.87

ta

0.79

 Nanto

0.78

Activations Density 0.117%

concepts related to security and caution

No Comments

No Known Activations