INDEX

Explanations

social and political terms and concepts

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

KD

-0.60

 TEAM

-0.60

 Socket

-0.59

 interviewer

-0.56

 Veterinary

-0.56

 referen

-0.56

 freezer

-0.55

 diabetic

-0.54

 Stab

-0.54

lined

-0.54

POSITIVE LOGITS

 itself

1.01

anship

0.95

lessness

0.93

ism

0.92

ophobia

0.87

atism

0.86

utics

0.86

 everywhere

0.85

liness

0.84

icity

0.83

Activations Density 1.727%

social and political terms and concepts

No Comments

No Known Activations