INDEX

Explanations

mentions of political figures and government actions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

EStreamFrame

-0.72

ebook

-0.69

plain

-0.68

pus

-0.68

inventoryQuantity

-0.67

omez

-0.62

 guiActiveUn

-0.62

ucci

-0.60

ajo

-0.60

plates

-0.60

POSITIVE LOGITS

 States

1.50

 Nations

1.29

 STATES

1.17

 Kingdom

1.10

States

1.04

 Methodist

1.03

 Federation

0.94

 State

0.85

 Airlines

0.80

 Nation

0.80

Activations Density 2.505%

mentions of political figures and government actions

No Comments

No Known Activations