INDEX

Explanations

news headlines or reports about unusual or noteworthy events

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 envy

-0.75

galitarian

-0.74

opian

-0.72

 passionately

-0.71

 hypocritical

-0.71

 caricature

-0.70

reprene

-0.69

 fantas

-0.69

obook

-0.69

 shameless

-0.68

POSITIVE LOGITS

 crews

1.00

 personnel

0.97

 evacuated

0.95

Initial

0.93

 precaution

0.91

 preliminary

0.90

Officers

0.87

 reopened

0.86

Officials

0.86

 cleanup

0.86

Activations Density 1.001%

news headlines or reports about unusual or noteworthy events

No Comments

No Known Activations