INDEX

Explanations

phrases related to expressing strong opinions or urging action

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

mith

-0.65

emale

-0.64

cffffcc

-0.64

 Mehran

-0.62

"}],"

-0.62

 constitu

-0.61

 wheelchair

-0.58

]=

-0.56

chief

-0.56

ometown

-0.56

POSITIVE LOGITS

 yourselves

1.28

 yourself

1.23

 Yourself

1.23

 your

0.97

 Your

0.94

 Them

0.87

ings

0.86

ership

0.86

tainment

0.82

ments

0.81

Activations Density 1.170%

phrases related to expressing strong opinions or urging action

No Comments

No Known Activations