INDEX

Explanations

phrases that indicate discrimination or biased judgments based on various criteria such as gender or race

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

uckland

-0.85

hiba

-0.74

jet

-0.70

along

-0.69

soon

-0.68

Edit

-0.67

Jet

-0.64

jam

-0.64

bats

-0.64

nin

-0.63

POSITIVE LOGITS

 nationality

1.04

 ethnicity

0.99

 gender

0.90

 sheer

0.90

 resemblance

0.88

 conscience

0.86

 merit

0.86

 whim

0.85

 principles

0.85

 disability

0.82

Activations Density 0.075%

phrases that indicate discrimination or biased judgments based on various criteria such as gender or race

No Comments

No Known Activations