INDEX

Explanations

content that denotes negativity or disapproval

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

 Fiscal

-0.80

ND

-0.78

 Prob

-0.77

MEN

-0.73

 Monetary

-0.73

 Expend

-0.73

 Feder

-0.72

 Confederation

-0.71

 Emirates

-0.70

Soc

-0.69

POSITIVE LOGITS

inducing

1.54

esque

1.52

themed

1.51

style

1.50

shaped

1.47

like

1.46

inspired

1.45

covered

1.43

colored

1.37

filled

1.36

Activations Density 0.062%

content that denotes negativity or disapproval

No Comments

No Known Activations