INDEX

Explanations

phrases related to spreading information, specifically encouraging others to share and promote content

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

deen

-0.80

*/(

-0.73

Zup

-0.72

clamation

-0.69

herty

-0.66

--+

-0.65

tarians

-0.64

venge

-0.64

 Starship

-0.62

udeau

-0.62

POSITIVE LOGITS

sheets

1.84

sheet

1.38

 misinformation

0.93

shirt

0.88

 disinformation

0.82

 awareness

0.82

 geographically

0.80

 across

0.80

 spreads

0.80

pread

0.78

Activations Density 0.036%

phrases related to spreading information, specifically encouraging others to share and promote content

No Comments

No Known Activations