INDEX

Explanations

computer programming terms related to data manipulation and analysis

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

 fuels

-0.27

anguages

-0.27

 empires

-0.27

pires

-0.23

atform

-0.23

deals

-0.22

 crises

-0.22

conom

-0.22

 riches

-0.21

rencies

-0.21

POSITIVE LOGITS

 referen

0.28

tein

0.28

yip

0.28

nesday

0.24

sic

0.24

tho

0.24

 reference

0.24

nen

0.24

[(

0.24

 eleph

0.24

Activations Density 19.645%

computer programming terms related to data manipulation and analysis

No Comments

No Known Activations