INDEX

Explanations

associations with the word "as."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

abase

-0.74

ndra

-0.71

mint

-0.67

arth

-0.64

toc

-0.62

achus

-0.62

bly

-0.60

 fray

-0.59

rences

-0.59

nexus

-0.58

POSITIVE LOGITS

 Sabha

0.93

Constructed

0.79

ä½ľ

0.70

 Devils

0.68

Ll

0.66

illard

0.64

ĳå£«

0.61

CRIP

0.60

à¨

0.60

Jose

0.60

Activations Density 0.094%

associations with the word "as."

No Comments

No Known Activations