INDEX

Explanations

phrases related to politics, government, and public statements

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

 namely

-0.72

 suppose

-0.72

gat

-0.69

 according

-0.67

leeve

-0.66

 whenever

-0.65

rand

-0.64

 owing

-0.64

-0.63

 representing

-0.63

POSITIVE LOGITS

 same

1.60

 slightest

1.56

 entirety

1.38

 requisite

1.31

 smallest

1.29

 brunt

1.27

 latter

1.25

 highest

1.23

 toughest

1.23

 simplest

1.21

Activations Density 2.680%

phrases related to politics, government, and public statements

No Comments

No Known Activations