INDEX

Explanations

phrases related to software or technology features

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

my

-0.66

ican

-0.64

 Leilan

-0.64

Ĥİ

-0.64

dos

-0.62

 Shooting

-0.61

 Shooter

-0.61

Published

-0.60

 Cruel

-0.60

 Rollins

-0.60

POSITIVE LOGITS

built

0.78

 atop

0.77

 upon

0.76

ILD

0.74

 into

0.72

 bridges

0.71

otypes

0.71

itect

0.70

packs

0.70

up

0.67

Activations Density 0.023%

phrases related to software or technology features

No Comments

No Known Activations