INDEX

Explanations

proper nouns and names

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

cers

-0.76

 foremost

-0.70

ACTED

-0.70

Interstitial

-0.68

 minded

-0.63

riott

-0.63

nces

-0.60

 Lauder

-0.60

 Scots

-0.59

 ransom

-0.58

POSITIVE LOGITS

ombie

1.40

ebra

1.24

ombies

1.22

arro

1.09

odiac

1.09

iggurat

1.05

ealous

1.01

hou

1.00

eros

1.00

ymes

0.99

Activations Density 3.849%

proper nouns and names

No Comments

No Known Activations