INDEX

Explanations

references to historical and political events or figures

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

sburgh

-0.71

hetti

-0.66

inventoryQuantity

-0.64

 Mara

-0.64

ocene

-0.63

ships

-0.63

itudes

-0.62

akia

-0.62

][/

-0.61

 Haley

-0.60

POSITIVE LOGITS

ynamic

0.88

cember

0.86

ensional

0.79

minist

0.78

ynam

0.77

hyde

0.76

ministic

0.76

angler

0.74

etermin

0.74

agram

0.73

Activations Density 2.489%

references to historical and political events or figures

No Comments

No Known Activations