INDEX

Explanations

terms expressing degrees of approximation or frequency

oai_token-act-pair · gpt-4o-mini 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

 redistributed

-0.62

ned

-0.60

ells

-0.60

 jaws

-0.56

 urge

-0.56

ella

-0.56

Memory

-0.55

osterone

-0.54

 tallest

-0.53

 enthus

-0.53

POSITIVE LOGITS

 awhile

0.91

 eternity

0.79

 millennia

0.77

 decades

0.75

 YEAR

0.71

 hours

0.71

 centuries

0.68

 millennium

0.68

hours

0.68

months

0.67

Activations Density 0.049%

terms expressing degrees of approximation or frequency

No Comments

No Known Activations