INDEX

Explanations

phrases related to problems or concerns

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

DOS

-0.85

ortex

-0.78

ude

-0.72

inders

-0.72

phones

-0.70

ilt

-0.69

rons

-0.66

oise

-0.66

de

-0.65

ardi

-0.65

POSITIVE LOGITS

 else

1.56

Else

1.49

 resembling

1.11

 Else

1.01

 intangible

0.96

 unforeseen

0.96

else

0.94

 akin

0.93

 miraculous

0.91

 unexpected

0.90

Activations Density 0.382%

phrases related to problems or concerns

No Comments

No Known Activations