INDEX

Explanations

references to media coverage and news reporting

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

 Reconstruction

-0.68

 Reson

-0.66

inho

-0.65

cker

-0.65

owsky

-0.64

ranged

-0.63

 Chili

-0.63

 Parenthood

-0.63

combe

-0.61

 Neighborhood

-0.61

POSITIVE LOGITS

eval

1.44

 outlets

1.10

 corps

0.96

mog

0.93

 outlet

0.92

 coverage

0.89

urized

0.88

 mogul

0.83

 circus

0.81

 conferences

0.80

Activations Density 1.385%

references to media coverage and news reporting

No Comments

No Known Activations