INDEX

Explanations

phrases related to blog posts and current events

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

CTR

-0.70

 Dickinson

-0.70

ãĥ¼ãĥĨãĤ£

-0.64

 Sergeant

-0.63

CoC

-0.63

PLA

-0.63

yards

-0.63

terday

-0.61

 Arist

-0.60

idential

-0.60

POSITIVE LOGITS

ceans

1.30

vernight

1.21

scill

1.16

atmeal

1.15

mbudsman

1.14

culus

1.12

lymp

1.11

phthal

1.03

mega

1.03

zzy

1.02

Activations Density 0.456%

phrases related to blog posts and current events

No Comments

No Known Activations