INDEX

Explanations

words related to criminal activity and law enforcement actions

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

anan

-0.76

 Canal

-0.75

 Startup

-0.75

 Horizon

-0.69

 Assignment

-0.68

 Tropical

-0.66

payer

-0.65

é¾įåĸļå£«

-0.65

 Improvement

-0.65

soType

-0.64

POSITIVE LOGITS

 lurking

1.08

 vanquished

1.06

 outnumbered

0.99

 attacking

0.97

 fleeing

0.96

 prey

0.95

 plotting

0.95

 exploiting

0.95

 hunted

0.94

 prow

0.94

Activations Density 0.296%

words related to criminal activity and law enforcement actions

No Comments

No Known Activations