INDEX

Explanations

phrases related to politics and public policy

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

isson

-0.76

aram

-0.75

hari

-0.74

ptions

-0.71

 assumes

-0.69

tec

-0.67

vet

-0.67

fitted

-0.67

CHAPTER

-0.67

fman

-0.66

POSITIVE LOGITS

 sake

1.22

 foreseeable

1.10

 entire

1.09

 rest

1.03

 nation

1.02

 remainder

1.02

 fledgling

1.01

 wearer

0.97

 country

0.94

 poorest

0.92

Activations Density 0.223%

phrases related to politics and public policy

No Comments

No Known Activations