INDEX

Explanations

political terms and entities, particularly related to the Democratic Party

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

obyl

-0.77

contact

-0.76

ador

-0.69

andise

-0.67

iceps

-0.66

 Drawn

-0.66

ĸļ

-0.66

phot

-0.66

hing

-0.66

obo

-0.65

POSITIVE LOGITS

 Party

1.26

 stronghold

1.09

 orthodoxy

1.08

 strongh

1.08

 dogma

1.03

 leaning

1.00

 caucus

0.99

 operative

0.97

 frontrunner

0.96

 manifesto

0.95

Activations Density 3.287%

political terms and entities, particularly related to the Democratic Party

No Comments

No Known Activations