INDEX

Explanations

names related to popular culture, particularly the name "Michael"

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

erent

-0.80

hips

-0.75

ancial

-0.72

ought

-0.69

Interstitial

-0.67

*/(

-0.67

Republic

-0.65

\\\\\\\\

-0.65

ulum

-0.64

aways

-0.64

POSITIVE LOGITS

angelo

1.22

 Vick

0.89

Bie

0.89

 Angelo

0.86

 Archangel

0.85

 Flynn

0.83

 Hayden

0.82

 Phelps

0.81

 Hastings

0.80

 Bloomberg

0.78

Activations Density 0.431%

names related to popular culture, particularly the name "Michael"

No Comments

No Known Activations