INDEX

Explanations

phrases related to fairness and justice

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

apse

-0.86

CHAT

-0.83

hent

-0.76

OPLE

-0.74

Assembly

-0.68

uality

-0.68

acid

-0.65

OUS

-0.63

artifacts

-0.62

hal

-0.61

POSITIVE LOGITS

yt

1.15

grounds

1.07

fair

1.02

itably

0.88

iciary

0.87

ground

0.85

 compensation

0.78

child

0.77

trade

0.75

 fair

0.73

Activations Density 0.642%

phrases related to fairness and justice

No Comments

No Known Activations