INDEX

Explanations

mentions of significant numerical figures or statistics

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

jri

-0.79

beh

-0.76

milo

-0.73

ional

-0.70

urally

-0.70

bably

-0.70

 seiz

-0.69

 elig

-0.69

wagen

-0.68

smanship

-0.68

POSITIVE LOGITS

 Expand

1.25

 Close

0.82

CLOSE

0.80

 Unlock

0.80

 Unable

0.77

 Copyright

0.75

License

0.75

WATCH

0.71

Govern

0.71

 Submit

0.71

Activations Density 0.002%

mentions of significant numerical figures or statistics

No Comments

No Known Activations