INDEX

Explanations

phrases related to political statements or positions

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

Mania

-0.67

CLOSE

-0.67

bour

-0.62

 Niet

-0.61

assis

-0.60

Fram

-0.59

 Analy

-0.58

 Winc

-0.58

Mask

-0.57

phrine

-0.57

POSITIVE LOGITS

 sure

1.15

 strides

0.94

 excuses

0.91

hift

0.89

 landfall

0.88

 decisions

0.87

 headlines

0.86

ailable

0.83

 adjustments

0.82

 arrangements

0.82

Activations Density 0.124%

phrases related to political statements or positions

No Comments

No Known Activations