INDEX

Explanations

proper names, particularly the name "Andrew"

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

pmwiki

-0.94

naire

-0.83

ioned

-0.82

kies

-0.80

leaders

-0.80

hips

-0.80

stice

-0.78

PLA

-0.76

vana

-0.74

doors

-0.73

POSITIVE LOGITS

 Mellon

0.86

 Sachs

0.85

Ng

0.84

 Dice

0.83

 Gerr

0.82

 McMahon

0.82

 Wiggins

0.82

 Morton

0.80

 Jackson

0.79

 McCabe

0.79

Activations Density 0.014%

proper names, particularly the name "Andrew"

No Comments

No Known Activations