INDEX

Explanations

mentions of military-related terms and actions

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

âĶģ

-0.86

 Seller

-0.78

flush

-0.74

apest

-0.74

luck

-0.70

nob

-0.69

 Myster

-0.66

gger

-0.65

Publisher

-0.65

Lot

-0.65

POSITIVE LOGITS

 fatig

1.25

 barracks

1.16

 stationed

1.14

 commanders

1.08

 colonel

1.07

 commander

1.00

 Corps

1.00

 acad

0.97

 personnel

0.96

 garrison

0.96

Activations Density 0.757%

mentions of military-related terms and actions

No Comments

No Known Activations