INDEX

Explanations

mentions of news, politics, events, and tech-related topics

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

yrinth

-0.67

ija

-0.67

ajor

-0.67

 Baldwin

-0.65

ATIONAL

-0.65

 Lauder

-0.64

 Interstellar

-0.64

cci

-0.64

ABE

-0.63

_-

-0.63

POSITIVE LOGITS

most

1.21

deck

1.08

mast

1.06

ographical

1.05

 drawer

0.99

boards

0.99

ography

0.99

eka

0.89

coat

0.89

board

0.86

Activations Density 3.682%

mentions of news, politics, events, and tech-related topics

No Comments

No Known Activations