INDEX

Explanations

phrases related to collecting information or data

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

forth

-0.71

WAY

-0.65

croft

-0.63

way

-0.61

Leaks

-0.60

word

-0.60

mia

-0.59

 Herrera

-0.59

blind

-0.59

void

-0.59

POSITIVE LOGITS

 signatures

0.94

ivist

0.76

STON

0.76

ources

0.75

DragonMagazine

0.75

sonian

0.75

ennial

0.74

 collected

0.74

 gathered

0.71

ibles

0.71

Activations Density 0.892%

phrases related to collecting information or data

No Comments

No Known Activations