INDEX

Explanations

phrases related to political and controversial 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

creen

-0.82

ABE

-0.71

 Tasman

-0.70

 Belg

-0.69

wagen

-0.68

 shroud

-0.67

 Mirage

-0.67

iewicz

-0.67

 destro

-0.66

 Reprodu

-0.64

POSITIVE LOGITS

ª

1.33

ł

1.27

Ĳ

1.24

ĳ

1.17

¹

1.10

ı

1.08

Ĵ

1.07

«

1.02

¤

1.01

ľ

1.01

Activations Density 0.785%

phrases related to political and controversial statements

No Comments

No Known Activations