INDEX

Explanations

instances of the word "working" followed by other words

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

antha

-0.82

 Ukrain

-0.77

ylon

-0.73

 Bubble

-0.73

iren

-0.70

wcs

-0.70

etsk

-0.69

emonic

-0.67

 Augustus

-0.67

 constitu

-0.67

POSITIVE LOGITS

bench

1.23

 ethic

1.20

hops

1.07

station

1.06

flows

1.05

aday

1.02

 tirelessly

1.01

horse

0.99

forces

0.98

 collabor

0.97

Activations Density 1.139%

instances of the word "working" followed by other words

No Comments

No Known Activations