INDEX

Explanations

various forms of punctuation or dashes indicating breaks in thought or new sections

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

pex

-0.69

elfth

-0.60

abee

-0.56

ene

-0.55

ilee

-0.53

 advoc

-0.53

pping

-0.53

ji

-0.52

ruck

-0.52

 nuts

-0.50

POSITIVE LOGITS

————

0.98

————————

0.89

_-

0.83

feat

0.76

 namely

0.76

including

0.69

whether

0.66

please

0.66

photos

0.65

––

0.64

Activations Density 0.065%

various forms of punctuation or dashes indicating breaks in thought or new sections

No Comments

No Known Activations