INDEX

Explanations

numbers in a specific context or setting

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

asta

-0.75

geist

-0.72

vantage

-0.69

lance

-0.67

IAS

-0.66

actionDate

-0.64

ifice

-0.64

helps

-0.64

 Caption

-0.63

activity

-0.63

POSITIVE LOGITS

teenth

1.69

teen

1.59

 hundred

1.47

fold

1.33

 thousand

1.33

 dozen

1.27

eenth

1.17

 weeks

1.14

 thirds

1.12

 consecutive

1.12

Activations Density 1.938%

numbers in a specific context or setting

No Comments

No Known Activations