INDEX

Explanations

places, political entities, and activities related to government and international relations

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

illac

-0.79

atown

-0.72

avorable

-0.72

hner

-0.70

iors

-0.70

apego

-0.68

¥µ

-0.67

eah

-0.67

agonist

-0.67

adena

-0.66

POSITIVE LOGITS

Â£

1.25

 Parliament

1.05

 Â£

0.96

Britain

0.94

 Labour

0.93

Labour

0.91

 Farage

0.90

 Columb

0.89

 Scotland

0.88

 Gael

0.88

Activations Density 4.822%

places, political entities, and activities related to government and international relations

No Comments

No Known Activations