INDEX

Explanations

references to group identities or classifications

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

atform

-0.77

oire

-0.77

uncture

-0.75

ebin

-0.73

berra

-0.72

 landfall

-0.71

enance

-0.71

fax

-0.71

culosis

-0.70

ossession

-0.69

POSITIVE LOGITS

 slaves

1.22

 losers

1.22

 heroes

1.16

 replacements

1.16

 idiots

1.15

 pioneers

1.15

 ambassadors

1.14

 selves

1.14

 combatants

1.13

 criminals

1.13

Activations Density 0.235%

references to group identities or classifications

No Comments

No Known Activations