INDEX

Explanations

mentions of political events, figures, and controversies

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

»Ĵ

-0.76

idious

-0.75

ONES

-0.70

_>

-0.66

 garg

-0.66

%%%%

-0.64

 mercury

-0.64

bane

-0.63

oned

-0.62

utic

-0.61

POSITIVE LOGITS

 Convention

1.08

 Committee

1.04

 Conference

0.91

 Headquarters

0.90

 Commission

0.90

 Colleg

0.89

 Coalition

0.88

 Voter

0.85

hester

0.85

 Council

0.84

Activations Density 0.016%

mentions of political events, figures, and controversies

No Comments

No Known Activations