INDEX

Explanations

phrases related to fires or physical destruction

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

xus

-0.97

 Lans

-0.74

DonaldTrump

-0.68

Birth

-0.67

 Vide

-0.66

 Freed

-0.66

Vec

-0.65

edy

-0.64

 cannabin

-0.63

 Pict

-0.63

POSITIVE LOGITS

flies

1.23

storm

1.15

 exting

1.12

balls

1.09

storms

1.08

fly

1.05

fighting

0.98

brand

0.98

proof

0.97

trap

0.96

Activations Density 0.706%

phrases related to fires or physical destruction

No Comments

No Known Activations