INDEX

Explanations

keywords related to news events and political figures

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

dit

-0.86

lishes

-0.80

 facult

-0.73

 contrad

-0.72

Companies

-0.71

Medium

-0.71

SHIP

-0.71

 Applic

-0.69

reau

-0.69

iculture

-0.69

POSITIVE LOGITS

 massacre

1.35

 bombings

1.21

 bombing

1.16

 Massacre

1.16

 riots

1.15

 atroc

1.14

 shootings

1.12

 tragedy

1.11

 killings

1.07

ghazi

1.06

Activations Density 2.921%

keywords related to news events and political figures

No Comments

No Known Activations