INDEX

Explanations

unauthorized activities or terms related to technical issues and incidents

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

minster

-0.65

BALL

-0.60

senal

-0.59

Graham

-0.57

estones

-0.56

PLA

-0.56

 Nanto

-0.56

Cra

-0.56

 Christy

-0.56

frey

-0.56

POSITIVE LOGITS

akable

1.15

imensional

0.98

gment

0.95

ailable

0.91

ifiable

0.86

ailability

0.82

xual

0.81

cedented

0.80

ivable

0.80

orable

0.80

Activations Density 1.062%

unauthorized activities or terms related to technical issues and incidents

No Comments

No Known Activations