INDEX

Explanations

physical locations or structures

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

PDATED

-0.62

 qualitative

-0.60

 mathemat

-0.58

 cumulative

-0.55

PDATE

-0.55

AQ

-0.55

 referen

-0.53

 Versus

-0.52

ensable

-0.52

 Weeks

-0.52

POSITIVE LOGITS

 tray

0.79

pipe

0.75

cloth

0.74

 mound

0.71

 ledge

0.70

hole

0.69

 whilst

0.69

 pedest

0.68

boards

0.67

 doorway

0.67

Activations Density 0.268%

physical locations or structures

No Comments

No Known Activations