INDEX

Explanations

terms related to reimagining or recreating ideas and concepts

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

DragonMagazine

-0.74

asma

-0.70

 families

-0.67

ician

-0.65

 Immunity

-0.65

 contacts

-0.65

 slightest

-0.65

 claimant

-0.64

 hypoc

-0.64

 streaks

-0.62

POSITIVE LOGITS

plementation

0.95

ining

0.92

ple

0.91

vised

0.90

plement

0.90

position

0.88

posing

0.86

orient

0.85

igor

0.85

agine

0.84

Activations Density 0.004%

terms related to reimagining or recreating ideas and concepts

No Comments

No Known Activations