INDEX

Explanations

percentage values

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

kefeller

-0.70

ITED

-0.69

flies

-0.68

Rav

-0.68

Attributes

-0.64

IRT

-0.62

GROUP

-0.62

ATHER

-0.62

æ³

-0.62

anca

-0.60

POSITIVE LOGITS

imet

0.93

ABV

0.92

imeter

0.86

xual

0.85

iles

0.81

 apiece

0.75

 utilization

0.72

 increments

0.70

bps

0.70

uary

0.69

Activations Density 0.363%

percentage values

No Comments

No Known Activations