INDEX

Explanations

names of people or characters in textual contexts

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

ï¸ı

-0.76

Ö¼

-0.72

suit

-0.70

suits

-0.66

 WATCHED

-0.66

sburgh

-0.65

 guiIcon

-0.63

hovah

-0.63

uably

-0.59

llers

-0.59

POSITIVE LOGITS

igans

1.41

thus

1.27

ning

1.26

ufact

1.23

onymous

1.18

xiety

1.15

thood

1.14

alyst

1.13

omaly

1.11

nery

1.08

Activations Density 0.855%

names of people or characters in textual contexts

No Comments

No Known Activations