INDEX

Explanations

acronyms or phrases that start with 'A'

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

enegger

-0.68

 optics

-0.63

 upstairs

-0.59

reports

-0.58

 noses

-0.58

IPS

-0.58

 eSports

-0.58

 XCOM

-0.56

falls

-0.55

 Indigo

-0.55

POSITIVE LOGITS

cknow

1.30

perture

1.21

ussie

1.19

BILITIES

1.19

verages

1.18

cknowled

1.12

BILITY

1.09

uctions

1.06

ryan

1.05

ileen

1.04

Activations Density 0.131%

acronyms or phrases that start with 'A'

No Comments

No Known Activations