INDEX

Explanations

names of countries, politicians, political parties, and related terms

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

agy

-0.70

DonaldTrump

-0.70

repre

-0.66

outer

-0.65

 Doodle

-0.62

Lens

-0.62

Magikarp

-0.61

 accur

-0.61

arse

-0.60

ppings

-0.59

POSITIVE LOGITS

IRO

0.86

 Turnbull

0.85

é¾

0.82

elong

0.79

DEF

0.74

 Tasmania

0.74

uckland

0.73

flix

0.72

icum

0.72

isher

0.71

Activations Density 2.786%

names of countries, politicians, political parties, and related terms

No Comments

No Known Activations