INDEX

Explanations

instances of the word "working" followed by different context

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

-0.79

 Ukrain

-0.77

 Bubble

-0.72

 Flavoring

-0.69

 constitu

-0.68

 Augustus

-0.67

Ń·

-0.67

anamo

-0.67

ylon

-0.66

emonic

-0.66

POSITIVE LOGITS

bench

1.21

 ethic

1.20

aday

1.09

station

1.09

flows

1.08

hops

1.07

horse

1.01

forces

1.00

 overtime

0.98

heet

0.97

Activations Density 3.156%

instances of the word "working" followed by different context

No Comments

No Known Activations