INDEX

Explanations

phrases related to text formatting

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

doms

-0.86

roma

-0.83

worth

-0.74

nee

-0.74

guard

-0.73

arma

-0.73

ĺħ

-0.72

hiro

-0.69

minent

-0.68

armac

-0.68

POSITIVE LOGITS

ters

1.00

ting

0.87

 format

0.87

atted

0.81

tered

0.77

 formats

0.77

aldehyde

0.75

 Format

0.74

etter

0.73

tering

0.73

Activations Density 0.021%

phrases related to text formatting

No Comments

No Known Activations