INDEX

Explanations

the German word "von" and its variations, indicating nobility or lineage

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

Interstitial

-0.81

taboola

-0.80

thumbnails

-0.69

gallery

-0.68

orative

-0.68

à¤¾

-0.68

inals

-0.67

inates

-0.67

Canadian

-0.64

âĶģ

-0.63

POSITIVE LOGITS

 Braun

0.99

 Frey

0.90

der

0.81

neg

0.80

env

0.78

heim

0.76

 Schwarz

0.76

 Doom

0.76

wald

0.73

 Britann

0.72

Activations Density 0.006%

the German word "von" and its variations, indicating nobility or lineage

No Comments

No Known Activations