INDEX

Explanations

ads or advertisement-related phrases

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

fruit

-0.75

ĵĺ

-0.74

terday

-0.70

Ago

-0.67

 pity

-0.66

¬¼

-0.62

 Sunshine

-0.61

 Stras

-0.60

chnology

-0.60

 Constantin

-0.60

POSITIVE LOGITS

rill

1.31

ouble

1.28

irect

1.26

icts

1.25

ragon

1.25

itions

1.24

der

1.20

iamond

1.19

aily

1.18

ifferent

1.17

Activations Density 2.018%

ads or advertisement-related phrases

No Comments

No Known Activations