INDEX

Explanations

numeric values or reference numbers in various contexts

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

onite

-0.80

reon

-0.79

 reckoning

-0.78

izons

-0.78

bered

-0.74

inement

-0.73

 underwater

-0.70

borg

-0.69

pez

-0.69

isconsin

-0.68

POSITIVE LOGITS

 Originally

1.11

 Contrary

1.03

 Exactly

1.03

 Whilst

1.02

 Introduced

0.96

 Various

0.96

 While

0.96

 Essentially

0.95

 Basically

0.94

 Whenever

0.94

Activations Density 0.047%

numeric values or reference numbers in various contexts

No Comments

No Known Activations