INDEX

Explanations

words related to political and corporate dynamics or relationships

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

 decomp

-0.72

 JPEG

-0.68

gib

-0.65

 silhou

-0.63

"+

-0.62

+++

-0.61

Kiw

-0.60

 Axel

-0.58

 veget

-0.57

 hurd

-0.56

POSITIVE LOGITS

1.29

ses

1.05

ski

1.03

ship

1.00

tal

1.00

tis

0.98

ese

0.93

tarian

0.93

lis

0.92

Activations Density 0.219%

words related to political and corporate dynamics or relationships

No Comments

No Known Activations