INDEX

Explanations

proper names or identifiers, particularly related to individuals and their titles

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

holders

-0.82

-0.74

 redes

-0.73

SPONSORED

-0.70

dayName

-0.69

 heck

-0.68

 constit

-0.67

 tranquil

-0.65

holder

-0.65

 favour

-0.64

POSITIVE LOGITS

.,

1.13

.?

0.95

ealous

0.92

acket

0.92

.:

0.91

ONES

0.90

iggins

0.88

AMES

0.88

ocket

0.87

ugg

0.84

Activations Density 0.008%

proper names or identifiers, particularly related to individuals and their titles

No Comments

No Known Activations