INDEX

Explanations

termed abbreviations or acronyms

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

lift

-0.69

 lawy

-0.66

draw

-0.65

lez

-0.63

 strawberries

-0.59

BOOK

-0.58

 gallery

-0.57

 override

-0.57

 nexus

-0.57

ARDIS

-0.57

POSITIVE LOGITS

essage

1.39

useum

1.27

ilitary

1.23

obile

1.23

otor

1.21

icro

1.19

obiles

1.15

otive

1.13

agnetic

1.11

pty

1.10

Activations Density 0.998%

termed abbreviations or acronyms

No Comments

No Known Activations