INDEX

Explanations

phrases indicating possession or ownership

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

arter

-0.69

OOL

-0.65

 Apart

-0.65

hatt

-0.62

hand

-0.62

 Anyway

-0.61

aneous

-0.61

 Recap

-0.60

anners

-0.59

 guiName

-0.58

POSITIVE LOGITS

 been

1.39

been

1.23

 undergone

1.17

 gotten

1.01

 endured

1.01

 arisen

1.00

 become

0.98

 risen

0.98

 fallen

0.97

 traditionally

0.96

Activations Density 0.109%

phrases indicating possession or ownership

No Comments

No Known Activations