INDEX

Explanations

numbers

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rador

-0.82

ovie

-0.73

aimon

-0.70

 Despair

-0.69

 Directorate

-0.69

rait

-0.68

axter

-0.66

 appre

-0.66

ossier

-0.65

clair

-0.65

POSITIVE LOGITS

enance

0.99

plates

0.89

metry

0.86

 crunch

0.84

esses

0.79

encies

0.69

number

0.69

mable

0.69

pad

0.68

limits

0.68

Activations Density 1.471%

numbers

No Comments

No Known Activations